NBER WORKING PAPER SERIES

WHAT DOES AN ELECTRIC VEHICLE REPLACE?

Jianwei XingBenjamin Leard

Shanjun Li

Working Paper 25771http://www.nber.org/papers/w25771

NATIONAL BUREAU OF ECONOMIC RESEARCH1050 Massachusetts Avenue

Cambridge, MA 02138April 2019, Revised February 2021

The views expressed herein are those of the authors and do not necessarily reflect the views of the National Bureau of Economic Research.

NBER working papers are circulated for discussion and comment purposes. They have not been peer-reviewed or been subject to the review by the NBER Board of Directors that accompanies official NBER publications.

© 2019 by Jianwei Xing, Benjamin Leard, and Shanjun Li. All rights reserved. Short sections of text, not to exceed two paragraphs, may be quoted without explicit permission provided that full credit, including © notice, is given to the source.

What Does an Electric Vehicle Replace?Jianwei Xing, Benjamin Leard, and Shanjun Li NBER Working Paper No. 25771April 2019, Revised February 2021JEL No. Q4,Q48,Q55

ABSTRACT

The emissions reductions from the adoption of a new transportation technology depend on the emissions from the new technology relative to those from the displaced technology. We evaluate the emissions reductions from electric vehicles (EVs) by identifying which vehicles would have been purchased had EVs not been available. We do so by estimating a random coefficients discrete choice model of new vehicle demand and simulating counterfactual sales with EVs no longer subsidized or removed from the new vehicle market. Our results suggest that vehicles that EVs replace are relatively fuel-efficient: EVs replace gasoline vehicles with an average fuel economy of 4.2 mpg above the fleet-wide average and 12 percent of them replace hybrid vehicles. This implies that ignoring the non-random replacement of gasoline vehicles would result in overestimating emissions benefits of EVs by 39 percent. Federal income tax credits resulted in a 29 percent increase in EV sales, but 70 percent of the credits were obtained by households that would have bought an EV without the credits. By simulating alternative subsidy designs, we find that a subsidy designed to provide greater incentives to low-income households would have been more cost effective and less regressive.

Jianwei XingNational School of DevelopmentPeking UniversityNo.5 Yiheyuan RoadBeijing 100871Chinajerryxing@nsd.pku.edu.cn

Benjamin LeardResources for the Future1616 P St. NWWashington, DC 20036leard@rff.org

Shanjun LiCornell University405 Warren HallIthaca, NY 14853and NBERSL2448@cornell.edu

1 Introduction

The diffusion of plug-in hybrid and fully electric vehicles (EVs), coupled with cleaner electricity

generation, offers a promising pathway to reduce air pollution from on-road vehicles and to

strengthen energy security. In contrast to conventional gasoline vehicles with internal combustion

engines, EVs use electricity stored in rechargeable batteries to power the motor. When operated

in all-electric mode, EVs consume no gasoline and produce zero tailpipe emissions. But the stored

electricity is generated from other sources such as power plants, which produce air pollution.

Therefore, the environmental impacts of EVs depend on several critical factors. First, emissions

created from operating EVs depend on the fuel source of electricity generation. Second, emissions

diverted from EVs depend on the difference in emissions intensity between EVs and the vehicles

that EVs replace. Third, some other factors also affect the emission impacts of EVs, such as

their effect on total vehicles in operation and the total vehicle miles travelled. While prior

literature has focused on the first factor (Archsmith et al., 2015; Holland et al., 2016, 2019),

few analyses explore the others factors. In this study, we fill this gap by focusing on the second

factor and illustrate the critical role this channel plays in determining the environmental benefits

of EVs.1 More specifically, we examine what EV buyers would have purchased had EVs been

unavailable. This counterfactual serves as the proper baseline to evaluate the emissions impacts

of EV diffusion.

An EV is a vehicle that is capable of running on electricity along at least part of the miles

that the vehicle is driven. These vehicles include both plug-in hybrids (e.g. Chevrolet Volt)

and fully electric vehicles (e.g. Tesla Model S) but exclude conventional hybrid vehicles (e.g.

Toyota Prius). Since the introduction of the first mass-market models into the United States

in late 2010, EV sales have grown rapidly to about two percent of the new vehicle market.2.

To encourage adoption, the federal government provides a federal income tax credit to new EV

buyers based on each vehicle’s battery capacity and the gross vehicle weight rating, with the

amount ranging from $2,500 to $7,500. Several states have established additional state-level

incentives to further promote EV adoption, including tax exemptions and rebates for EVs and

non-monetary incentives such as high-occupancy vehicle (HOV) lane access, toll reduction, and

free parking.3

A potential concern associated with subsidy policies is that they may create “non-additional”

1Below in Section 7, we also discuss in more detail how the other two channels would affect our results.2See Appendix Table A.1 for the average annual market share.3In addition, federal, state, and local governments also provide funding to support charging station deployment.

1

emissions reductions: some EV buyers would have purchased EVs even if there was no subsidy.4

Since early adopters may place a higher value for new technology and the environment, it is

likely that some buyers have received a windfall gain without changing their behavior.

Moreover, even if the tax credits increased EV sales, the emissions impact may be small if

EVs replace vehicles with low emissions ratings. The effect that EVs have on emissions depends

on how clean EVs are relative to the vehicles they are replacing. Many EV buyers could have

bought a low-emission gasoline vehicle had EVs or EV incentives not been available. This could

arise from consumer preference heterogeneity and sorting: consumers that value fuel efficiency or

environmentally friendly vehicles buy vehicles that are fuel-efficient or deemed environmentally

friendly, such as the Toyota Prius or the Toyota Prius Plug-In. For these buyers, opting to buy

the EV yields small or even negative emissions benefits.

To understand these issues, we use a stylized model to derive a simple expression relating

vehicle substitution patterns – represented by cross-price elasticities of demand – to emissions

changes. Our model shows that the greater the substitution a non-EV has with an EV, the

greater the impact the vehicle’s emissions have on the emissions effect of the EV. We then

estimate a random coefficient discrete choice model of vehicle demand by using a rich household

survey of US new vehicle buyers and market-level sales data from 2010 to 2014. The estimation

takes advantage of the second-choice information from household survey data, which greatly

improves the precision of the random coefficient estimates and the resulting substitution patterns.

With the model, we simulate counterfactual market outcomes by removing the EVs from the

market to examine how consumers substitute between EVs and non-EVs. We then conduct

other counterfactual exercises to examine the cost-effectiveness of the income tax credits policy

in terms of reducing on-road emissions and compare these results with alternative policy designs.

Our approach builds on the methodology used by Holland et al. (2016) and Holland et al.

(2019) to estimate EV replacement vehicles. Their approach assigns a replacement vehicle based

on stated preference second choice survey data.5 Instead of using survey data solely to assign a

4Additionality is a key issue for many other subsidy policies such as carbon offset programs (Bento et al.,2015) and subsidy programs for alternative-fuel vehicles (Beresteanu and Li, 2011; Huse, 2014).

5Holland et al. (2016) create a composite substitute gasoline vehicle for each EV by taking the weighted averageof emissions of the top gasoline substitute vehicles reported in the survey. But they do not have substitute choicedata for certain EV models including the Honda Fit EV, Fiat 500 EV, and BYD e6. In addition, the approach inHolland et al. (2016) assumes that sales of a specific EV model replace the same gasoline vehicle, which might bestrong. For example, because of heterogeneous consumer preferences, some Nissan LEAFs replace a Toyota Prius,while other Nissan LEAFs might replace a Ford Fusion. We define theoretically the emissions of a compositevehicle that accurately represent the emissions of all vehicles that replace an EV. This definition is a weightedaverage of the emissions of all vehicles that are substitutes for an EV, where the weights are proportional to each

2

substitute model for each EV, we estimate a vehicle demand model incorporating both aggregate

sales data and second choice survey data. The estimated own- and cross-price elasticities can

directly reflect the substitution patterns between EVs and vehicles of other fuel types. The

recovered consumer preference parameters allow us to run simulations to quantify the difference

in emissions between the observed EV sales and the simulated replaced vehicles, as well as the

impact of the subsidy programs on increasing EV sales. Our structural approach also allows us

to compute how much EV subsidies lead to additional EV purchases, which allows us to evaluate

the cost-effectiveness of the subsidies.

With our estimated demand model, we run counterfactual simulations, which reveal three

key findings. First, electric vehicles appear to be replacing relatively fuel-efficient vehicles,

as households that generally prefer EVs also prefer conventional gasoline vehicles with better

fuel economy. This implies that calculations of emissions benefits should account for the non-

random replacement of conventional gasoline vehicles. Ignoring this non-random replacement

overestimates emissions benefits by 27 percent. Second, the availability of and support for EVs

has not led to a significant reduction in market share for hybrid vehicles. Hybrids had been

supported by the federal government in the 2000s and have seen a decline in market share since

2014, a time when EVs had started to gain significant market share. But our results suggest that

EVs have had a limited impact on hybrid sales. Instead, the elimination of the federal subsidy

for hybrids has caused a significant reduction in hybrid sales. Third, the cost-effectiveness of

the subsidy program is limited by the fact that about 70 percent of consumers would have

purchased EVs without the subsidy. We find that this result is sensitive to the price elasticity

of demand, where more elastic demand implies a greater number of additional EV purchases.

By comparing the current uniform subsidy with an alternative policy design that removes the

subsidy for high-income households and provides additional subsidies to low-income households,

our analysis shows that better targeting could potentially increase the cost-effectiveness of the

subsidy programs in terms of EV demand and environmental benefits. Our simulation results

contribute to the literature on the diffusion of low-emission technologies and the cost-effectiveness

of subsidy programs promoting these technologies (Allcott et al., 2015; Boomhower and Davis,

2014; Langer and Lemoine, 2018; Sallee, 2011).6

Our study adds to the literature on the demand for electric vehicles and the EV market.

Li et al. (2017) employ data on EV sales and charging stations at the city level to quantify

vehicle’s cross-price elasticity of demand with respect to the EV’s effective price.6See Appendix A for a detailed review of this literature.

3

the interplay between the availability of charging infrastructure and the installed base of EVs.

Our structural approach allows us to address several key issues surrounding EV demand that

reduced-form methods are unable to quantify, including the identification of vehicles that are

being replaced by EVs and the welfare effects of EV policies. Springel (2020) estimates a

structural model of consumer vehicle choice and charging station entry in the Norwegian EV

market and compares the effectiveness of direct purchasing price subsidies with charging station

subsidies. Li (2016) examines the issues of compatibility in charging technology and finds that

mandating compatibility in charging standards would increase the sales of EVs. Muehlegger and

Rapson (2020) use the EV subsidy receipts data and vehicle transaction prices to estimate the

pass-through rate of the EV incentive program in California and find that 80 to 92 percent of

the subsidies were passed through to consumers and that a decrease of 10 percent in EV prices

increases EV demand by 31 to 37 percent. In contrast to these papers, our study focuses on

identifying the vehicles that EVs replace.

We organize the rest of the paper as follows. Section 2 briefly describes the data used in the

empirical analysis. In Section 3, we develop a simple analytical model to show emissions impacts

of vehicle substitution depend on key vehicle demand parameters to help guide our empirical

analysis. Section 4 presents the empirical model and estimation strategy. Section 5 presents the

estimation results of the substitution. In Section 6, we present the counterfactual simulations

to evaluate the environmental benefits of the introduction of EVs and the impact of the EV

subsidy. We also conduct simulations to examine the impact of EVs on hybrid vehicle sales and

whether an income-dependent subsidy design could improve the cost-effectiveness of the subsidy.

Section 8 concludes.

2 Data Description

We use three data sets to estimate the model of vehicle demand. The primary data source is

household-level survey data from the US New Vehicle Customer Study by MaritzCX Research.

It is a monthly survey of households that purchased or leased new vehicles. The data provide

detailed information of demographic characteristics of households that purchased each vehicle,

and the alternative vehicles they considered while making the purchase decisions. We use survey

data for five model-years: model year (MY) 2010 through MY 2014, where each model year is

defined as September of the previous calendar year to August of the current calendar year. (For

example, MY 2011 is defined as September 2010-August 2011.) For computational purposes, we

4

draw a sample of 11,628 transactions from the data after removing observations with missing

observed consumer attributes or information on the purchased and seriously considered models,

and end up having 1,509, 1,860, 2,287, 2,899, and 3,073 transactions for MY 2010-MY 2014,

respectively.7 As the market share of EVs is tiny, so that would include enough EV observations

to have sufficient variation in consumer demographic attributes for EV buyers to identify the

preference for EVs among different demographics, we use non-random sampling by including

all EV observations from the survey sample and randomly drawing observations for the other

fuel types. To adjust for non-random sampling, we then follow Manski and Lerman (1977) to

include a weighted exogenous sample maximum likelihood by re-weighting each observation in

the likelihood. The weight is defined by the actual market share in the population divided by

the within-sample market share.

Table 1 summarizes the demographic information for the households that made those purchase

transactions. The average household income for the survey respondents in the sample is $140,448,

which is higher than the average household income of $117,795 for married couples in the United

States.8 This feature of the data is caused by oversampling consumers who purchased EVs and

hybrid vehicles. The average household size is 2.66 people, and 63.9 percent of the heads of

household have earned a college degree. Of the respondents, 66.1 percent of the respondents are

from an urban or suburban area, with an average commuting of 25.6 minutes and average gasoline

price of $3.48 during the survey time. About 50 percent of the sampled households selected a light

truck, and the average price of the vehicles that the sampled households purchased is $33,451.

The average fuel economy of the purchased vehicles is 34.8 mpg.9 Appendix Table A.3 provides

further descriptive statistics for new vehicle buyers by fuel type. EV buyers have a much higher

income and a larger percentage of them graduated with a college degree.

The household survey data also include alternative vehicle choices that consumers considered

while purchasing vehicles, providing a valuable source for identifying unobserved preference

heterogeneity. Table 2 summarizes the top alternative vehicle choices reported by survey

respondents for EV models. The data reflect that EV buyers have a strong preference for

alternative fuel technologies, since most of them still consider PHEVs or hybrid vehicles as their

7This sample is the maximum number of households that we can draw without running into memoryconstraints. It takes 10 hours to estimate the model with parallel processing. The estimation time risesproportionally with the sample size, i.e., doubling the sample size doubles the computation time.

8Data source: IRS Statistics of Income, 2014.9This average is significantly higher than the average fuel economy of all purchased vehicles during the sample

period because of the oversampling of EV and hybrid buyers.

5

second choices. This strong correlation of the fuel economy between the purchased vehicle and the

alternative choices greatly facilitates estimating the random coefficients for vehicle fuel economy.

For luxury EV models, such as Tesla Model S, customers might also consider luxury gasoline

models such as Audi A7 as their alternative choices. The proximity in price, size, and some other

observed vehicle attributes would help in identifying consumer heterogenous preference for those

attributes.

Figure 1 summarizes consumers’ second choices by fuel type based on the survey data and

reflects the heterogeneous preference of fuel type among different groups of consumers. Among

gasoline buyers, 96.9 percent would consider another gasoline vehicle as a second choice, 2.9%

percent would consider a hybrid vehicle model as an alternative, and only about 0.2 percent

would consider either BEVs or PHEVs as substitutes. Gasoline vehicle buyers, who are the

majority of new vehicle purchasers, are generally less interested in the EV technology. Hybrid

vehicle buyers demonstrate a stronger preference of fuel economy, and 39.7 percent of them would

consider another hybrid vehicle as an alternative choice. However, only 3 percent would consider

EVs as second choices. Those consumers who purchase hybrid vehicles enjoy vehicles that save

fuel cost but do not favor the plug-in feature of EVs. Both PHEV and BEV buyers show a strong

interest into EVs: many of them are considering another EV as their second choices. However,

34.5 percent of PHEV buyers consider a hybrid vehicle as an alternative, and only 16.5 percent

consider a BEV model. PHEV buyers are more willing to adopt the EV technology but are less

interested in all-electric vehicles, probably because of the limited range of BEVs. BEV buyers are

most into the EV technology, and 41.3 percent of them pick another BEV model as their second

choice, 25.3 percent consider PHEVs, and only 18.6 percent consider another gasoline vehicle as

substitute. BEV adopters are those who care most about the feature of electrification and those

who are most into the newest technologies. The general pattern of this figure reveals the strong

correlation between alternative choices and the purchased vehicles and reflects the critical role

that the substitution pattern plays in reflecting the heterogenous preference of consumers.

We merge vehicle characteristics data from Wards Automotive, which provide detailed

attributes of each vehicle model in each model year, including horsepower, size, curb weight,

wheelbase, and fuel economy.10 The data set is further complemented by aggregate vehicle sales

data, which provide market-level information on new vehicle demand, obtained from registration

data compiled by IHS Automotive. The IHS data record the quarterly number of registrations

10The Wards Automotive data have a fine level of vehicle identification detail. We merge base model year bymake, model and fuel type to the MaritzCX survey data, where the base model is defined as the trim with thelowest MSRP among all trims by make, model, and fuel type identification within the same model year.

6

for each new vehicle model, broken down by fuel type, which are aggregated to model-year level

to construct the market share for each new vehicle model in each model year. All the above

data sets are matched at the model year-make-model-fuel type level, for example, 2014 Ford

Focus gasoline, and the vehicle attributes are assigned using the base model. The total number

of vehicle models that are defined in the model-year choice sets are 424, 404, 418, 441, and 459

for MY 2010 - MY 2014, respectively. Table 1 summarizes the basic attributes of those vehicle

choices and the composition of the choice sets by fuel type.

We assign average vehicle prices based on respondent-reported transaction price information

in the MaritzCX survey data. Respondents are asked to report the sales or lease prices of their

vehicles, within a few months after purchase. These values reflect the price that households paid

on average for each vehicle and may be different from the traditionally used MSRPs because of

negotiations or temporary promotions. These prices exclude any credits received from trade-ins

and include sales taxes. We compute market by model by fuel type prices as the unweighted

average transaction price for all purchases and leases in the raw survey data. We do not adjust

these prices for tax credits or rebates because we do not observe whether households claimed

these incentives. Since many of the household observations lease plug-in electric or fully electric

vehicles, credits or rebates for these vehicles go to the leasing company, which then likely passes

through the incentive as a lower purchase price. We collect data of monthly average gasoline

prices by region from the Energy Information Administration (EIA).11 We convert all prices,

including average transaction prices and fuel costs, to real 2014 dollars using the Bureau of

Labor Statistics (BLS) Consumer Price Index.

We obtain detailed information on locations and open dates of all charging stations from the

Department of Energy’s Alternative Fuels Data Center (AFDC). By matching the zip code of

each charging station with the zip codes reported in the survey data, we assign the total number

of charging stations available in the city to each observed survey respondent.

3 Theory of Substitution

To motivate our empirical analysis, we lay out a stylized model of vehicle substitution to illustrate

how substitution between vehicles from a policy or non-policy change affects emissions. Consider

a new vehicle market where there are J unique models for sale, where each model is indexed by j.

11EIA reports monthly gasoline prices by region, defined by the Petroleum Administration for Defense Districts(PADDs). We assign gasoline prices to each sampled household based on its PADD region and the month ofvehicle purchase.

7

Model j has lifetime emissions equal to ej and has aggregate demand qj = qj(p1, p2, ..., pJ), where

pj represents the sales price net of subsidies for vehicle j. Total lifetime emissions of vehicles

sold are

E =J∑j=1

ejqj(p1, p2, ..., pJ). (1)

Without loss of generality, we assume that j = 1 is an EV and j = 2, 3., ..., J are gasoline or

hybrid models. We assume that the EV’s price is subsidized by an amount s, so that the EV’s

price is p1 = p01 − s, where p01 is the EV’s price without a subsidy. Differentiating total lifetime

emissions with respect to the EV subsidy yields

dE

ds= −e1

dq1dp1−

J∑j=2

ejdqjdp1

. (2)

Normalizing the change in the subsidy by the EV’s price (so that dEds′

= dEdsp1) and defining the

own-price and cross-price elasticity of demand with respect to the EV price as ε1 = dq1dp1

p1q1

and

εj =dqjdp1

p1qj

, respectively, we can express equation (2) as

dE

ds′= −e1q1ε1 −

J∑j=2

ejqjεj. (3)

Equation (3) reveals that the effect of the subsidy on lifetime emissions is proportional to the

own-price elasticity of demand for the EV and the cross-price elasticity of demands for all other

vehicles. The cross-price elasticities εj represent the substitution pattern between the EV and

the non-EV models. The larger the value of this derivative, the greater the substitution and the

more of an impact the non-EV model has on the emissions impact of the subsidy. Consider the

simple example where εj = 0 for j = 3, 4, ..., J . Then equation (3) becomes

dE

ds′= −e1q1ε1 − e2q2ε2. (4)

If the demand responses offset one another so that there is no change in total new vehicle sales,

then the change in emissions depends on the relative difference between the lifetime emissions of

the EV and the non-EV j = 2:dE

ds′= (e2 − e1)q1ε1. (5)

This simplified equation is conceptually the same approach taken by prior studies to quantify

8

the emissions impacts of EVs.

3.1 Defining a Composite Substitute

This approach above is an accurate representation of the full impact if (1) the only substitution

that takes place is between the EV and a single vehicle, (2) the single vehicle is the correct

substitute, or (3) the j = 2 model’s emissions accurately reflect the emissions of all the vehicles

that are substitutes for the EV. In most cases, an EV will have more than one vehicle as a

substitute. Here we derive a simple formula defining the emissions of a composite vehicle that

satisfies the third condition when more than one vehicle substitute for the EV. We begin by

assuming that a change in the subsidy does not change total vehicle sales: − dq1dp1

=J∑j=2

dqjdp1

. Denote

the emissions of the composite vehicle by ec. We want to find an ec that solves dEds′

= (ec−e1)q1ε1.Substituting this expression into equation (3) yields

dE

ds′= −e1q1ε1 −

J∑j=2

ejqjεj = (ec − e1)q1ε1. (6)

Making cancellations and isolating ec yields

ec = −J∑j=2

ejqjεjq1ε1

. (7)

Emissions for the composite vehicle equal to the product of emissions and the ratio of the

cross-price elasticity of demand are scaled by sales of vehicle j, and the own-price elasticity of

demand is scaled by sales of the EV. In our empirical demand model, we are able to identify

composite vehicle emissions based on estimated own-price and cross-price demand elasticities.

Equation (7) can be further simplified to

ec = −J∑j=2

ejdqjdq1

. (8)

This general expression can be used to accurately evaluate hypothetical settings where electric

vehicles are added or removed from the market. This expression suggests that evaluating the

impact of EV subsidy on reducing emissions depends on the estimation of the substitution pattern

between EVs and all the other vehicle models in the market.

9

3.2 Additionality

In this section, we derive an equation that shows how the non-additionality of a subsidy is affected

by demand parameters. We define non-additionality as the proportion of EVs that would have

been bought without the subsidy to the total EV sales with the subsidy. A higher ratio implies

more non-additional purchases and more subsidy dollars going to households that would have

bought an EV without the subsidy. The proportion is equal to

N =q1(p

01)

q1(p1). (9)

Differentiating N with respect to s′ and substituting the price elasticity of demand for the

EV yieldsdN

ds′=q1(p

01)

q1(p1)ε1 (10)

Evaluating equation (10) at the price where the subsidy is equal to zero (p01 = p1) yields

dN

ds′= ε1 (11)

This equation shows that non-additional purchases are proportional to the EV own-price

elasticity of demand. More elastic demand (more negative ε1) implies relatively fewer non-

additional purchases and a greater number of purchases that are created by the subsidy. In

contrast to the results we derived for the emissions impacts of the subsidy, the additionality of

the subsidy depends on the own-price elasticity of demand only.

4 Empirical Model and Estimation

In this section, we discuss our empirical model and estimation strategy. We estimate vehicle

demand preference parameters using a random coefficient discrete choice model in the spirit of

Berry et al. (1995, 2004), Petrin (2002), and Train and Winston (2007). Our model most closely

follows the structure of Train and Winston (2007), as we exploit household demographics and

second choice data to identify the model parameters.

10

4.1 Vehicle Demand

The household survey data are not representative of the entire population since they include

only buyers of new vehicles. Therefore, we model new vehicle preferences conditional on the

decision of buying a new vehicle. Our approach will not be able to capture the substitution

between the new vehicle models and the outside option: buying a used car, continuing using

the household’s old vehicle, or relying on public transportation. Instead, our model represents

how consumers choose among new vehicles and how changes in new vehicle attributes or the

selection of new vehicles available for purchase affects new vehicle sales. Two factors suggest

that our model could reasonably capture the substitution that consumers make when deciding

between an EV and another vehicle option. First, EVs represent a new segment of the light-duty

vehicle market, where few used vehicle options represent plausible substitutes. Second, EVs are

generally expensive options relative to most new or used vehicles. If consumers substitute among

similarly priced vehicles, the EV substitutes are likely to be expensive new vehicles.

We define household i’s utility from purchasing vehicle model j as:

uij =K∑k=1

xjkβk − α1lnpj + ξj︸ ︷︷ ︸δj

+α2lnpjYi

+∑kr

xjkzirβokr +

∑k

xjkvikβuk︸ ︷︷ ︸

µij

+εij,(12)

where δj is the mean utility of vehicle model j which is constant across consumers in the same

market. xjk stands for the kth vehicle attribute for model j. We include horsepower, weight,

gallons per mile12, and some vehicle segment dummy variables as the observed vehicle attributes.

Price pj is the average transaction price observed from the survey data, which is constant for the

same model by fuel type for all households buying a vehicle in the same market.13

The second component, µij, captures heterogeneous utility driven by both observed and

unobserved consumer characteristics. Yi is household i’s income in the corresponding year, and

we assume consumer price sensitivity to be inversely related to income. One would expect α2 to

be negative, as higher-income households would be less sensitive to a price increase because of

12For EVs, the gallons/mile measurement is defined as the inverse of miles/gallon gasoline equivalent, which isa metric defined by EPA to measure the energy efficiency of EVs.

13Averaging household-level transaction prices by vehicle alleviates recall bias at the household level and maybetter reflect actual price differences between different vehicle models. However, our construction of averagetransaction price may still introduce selection bias as transaction prices are reported by consumers who purchasedthat model. As a robustness check, we re-estimate the demand using MSRPs and the results are presented inAppendix Table A.4. The price coefficient and elasticity estimates are similar to our benchmark model that usesaverage transaction price.

11

the diminishing marginal utility of money. zir denotes consumer i’s other demographic variables-

including family size, education level, whether living in an urban area, the average gasoline

price, and the number of charging stations in the area- which are interacted with certain vehicle

attributes to capture variation in consumer preference due to observed heterogeneity.

The unobserved consumer taste vik is assumed to have a standard normal distribution. The

coefficient βuk can be interpreted as the standard deviation in the unobserved preference for

the vehicle attribute k conditional on the consumer’s observed attributes. Let θ = {βokr, βuk},denoting the “nonlinear” parameters, and it is understood that the vector δ = {δ1, ..., δj} is

estimated conditional on a given θ1. The last component, εij, is the idiosyncratic preference

of household i for vehicle model j, and it is assumed to have an i.i.d. type one extreme value

distribution.

A useful feature of the MaritzCX data is that they include vehicle models that consumers

seriously considered other than the purchased model. This allows for a ranking of both the first

and second vehicle choices.14 We exploit the second choice data as a source of variation to identify

unobserved heterogeneous preferences conditional on observed household characteristics.15 For

example, if the second choices of an EV model include only EV models, this would suggest that

EV buyers have a very strong preference for this particular fuel type. If, on the other hand,

the second choices include many non-EV counterparts of the EV models within the same make,

this would suggest a less strong preference for EV type, but preference for the same make is an

important factor. Similarly, the comparison between the chosen model and the second choice in

other dimensions of vehicle attributes such as vehicle size or fuel economy can also inform us

about consumer preference heterogeneity for these vehicle attributes.

To use the second choice information, we form the likelihood function based on the joint

probability of household i choosing j as the first choice and considering h as the second choice:

Pijh =

∫exp[δj(θ) + µij(θ)]∑g

exp[δg(θ) + µig(θ)]· exp[δh(θ) + µih(θ)]∑g 6=jexp[δg(θ) + µig(θ)]

f(v)dv. (13)

14We use survey response data from multiple questions to assign a second choice. The first question is “Whenshopping for your new vehicle, did you consider any other cars or trucks?” Respondents answering yes to thisquestion were then asked to provide make, model, model year, fuel type, and other vehicle information for themodel that they most seriously considered but did not purchase. Those second choices represent the alternativevehicle models that consumers considered when making their actual purchase decisions.

15Another benefit of incorporating second choice data is that they increase the efficiency of the estimation(Beresteanu and Zincenko, 2018).

12

The probability of observing household i choosing model j is conditional on the household’s

vi vector, and the probability is calculated by integrating over the distribution of v. Instead of

constructing moments exploiting the exogeneity assumption that unobserved product attributes

are uncorrelated with observed attributes, we use the maximum likelihood estimation (MLE)

method with a nested contraction mapping to estimate θ and δ (Train and Winston, 2007;

Langer, 2012; Goolsbee and Petrin, 2004; Whitefoot et al., 2013). Let lnRi = lnPijh, denoting the

individual log-likelihood of household i choosing the observed purchased model j and considering

the observed alternative choice h. The log-likelihood function of the entire sample for a single

market is therefore:

lnL =N∑i=1

lnRi. (14)

The nonlinear parameters θ are estimated by maximizing the likelihood function.16 Given

the larger number of mean utilities δ, we follow the two-step procedure in Berry et al. (1995),

which shows that under certain regularity conditions, for each θ, there exists a unique δ that

matches the predicted market shares with observed ones. The market demand is the sum of

individual consumers’ demand, and the predicted market share is calculated by calculating Pij

with parameters θ = {βokr, βuk} and δ = δ1, ..., δj and averaging over the N consumers in the

survey sample. Following this strategy, we back out the mean utility vector δ for any given θ

using the contraction mapping technique:

δtj(θ, S) = δt−1j (θ, S) + ln(Sj)− ln(Sj(θ, δt−1(θ, S))). (15)

We iterate this equation until the difference between δtj and δt−1j is smaller than a pre-specified

inner loop tolerance. We set a strict tolerance to be 10E-15 to reduce the likelihood of achieving

a local optimum. Once θ and δ are estimated using the MLE method, we then recover the

parameters in mean utility:

δj = −α1lnpj +K∑k=1

xjkβk + ξj,

where ξj denotes the unobserved vehicle attributes of model j. To control for the correlation

of price with the unobserved product attributes, following Train and Winston (2007), we use

BLP-style instruments that measure the sum of distance and squared distance in attribute space

between own product and other products in the same firm and from other firms.

16In Appendix B, we lay out more details of the likelihood function and the gradient for estimation.

13

4.2 Identification

Consumer utility is composed of three parts: mean utility, observed heterogeneity, and

unobserved heterogeneity. The linear parameters in the mean utility β and α1 are identified

through the variation in market shares corresponding to variation in price and other observed

vehicle attributes. Because of the potential correlation between price and the unobserved vehicle

attributes ξj, functions of attributes of other competing products that capture the intensity of

competition are used as instruments to provide exogenous variation in prices. The maintained

exogeneity assumption is that unobserved product attributes are not correlated with observed

product attributes.

The nonlinear parameters βokr and α2 in the observed individual heterogeneity component

are identified from the correlation between household demographics and vehicle attributes.

For example, if we observe that households with a high level of education disproportionately

purchased more electric vehicles, we would expect a positive coefficient for the interaction between

household education level and the EV dummy. If higher-income groups tend to be less sensitive

to vehicle prices and disproportionately buy more expensive vehicle models, we would expect a

negative sign for α2, which captures the impact of income on consumers’ price sensitivity.

The unobserved consumer heterogeneity parameters βuk are primarily identified by the

correlation between first and second choice vehicle attributes. For example, if consumers who

purchase high fuel-economy vehicles tend to state that they would have purchased a high fuel-

economy vehicle if their first choice was not available, we would expect a large coefficient for the

parameter associated with fuel costs (i.e., the standard deviation of the preference for the fuel

cost). Berry et al. (2004) note that having micro-level second-choice data helps the estimation of

random coefficients when they have observations for only one market year, and Train and Winston

(2007) also mention that including alternative choice data significantly improves the precision

of the random coefficient estimates. In contrast to these studies, however, the unobserved

heterogeneity parameters in our model are also identified by changes in choice sets over time.

We leverage the feature of our sample, which includes periods where few electric vehicles were

available (2010 and 2011), followed by periods of availability (2012-2014) and an expansion of

available options (see Appendix Table A.1). The variation in the choice sets over time provides

an additional source of identification for the random coefficients.

14

5 Estimation Results

We first report parameter estimates for the random-coefficient model and then use the estimates

to calculate price elasticities to show implied substitution patterns.

5.1 Parameter Estimates

Table 3 reports the estimation results of the demand model. The mean utility δ represents

the average preference consumers have for each vehicle model and is estimated by equating

the predicted market shares to the observed market shares. The mean preference coefficients

for price and each observed vehicle attribute are recovered from instrumental variables (IV)

estimation with the instruments accounting for the endogeneity of price. Both ordinary least

squares (OLS) and IV results are reported in panel (a) of Table 3 and reflect the preferences for

vehicle attributes that are generally expected. Consumers have a negative preference for price

and the price coefficient in the IV specification is more negative, suggesting OLS underestimates

the price sensitivity. Consumers have a positive preference for acceleration, measured by the

ratio of horsepower to weight, and also prefer heavier vehicles. The coefficient for gallons/mile

is positive but statistically insignificant as the operating cost of the vehicle ($/mile) has been

controlled in the consumer heterogeneity component. The sign for AFV dummy is negative,

suggesting consumers in general prefer conventional vehicles probably because of risk aversion,

range anxiety concerns and lack of awareness of AFVs and EVs. Conditional on other vehicle

attributes, consumers do not have a significantly different preference for pickup trucks relative

to passenger cars. The positive signs for MY 2011-MY 2014 dummies suggest that consumers

prefer vehicles in later model years relative to MY 2010, controlling for other vehicle attributes.

This is consistent with broad sales patterns during this time period, when total sales had been

recovering from the recession.

Turning to the consumer heterogeneity parameters, the interaction terms of consumer

demographics with vehicle attributes are estimated precisely with intuitive signs. The coefficient

of log(price) divided by income captures the extent to which a consumer’s price sensitivity varies

with income. The negative sign of the estimate suggests that households with lower income

react more negatively to a vehicle’s price than households with higher income. The elasticities

implied from the price preference are further discussed below. Households of a larger family

size prefer larger vehicles that are heavier. Compared with households that live in suburban

and rural areas, households that live in urban areas are less likely to adopt pickups, probably

15

because of less towing utility and limited parking space, but are more interested in EVs because

of both more frequent city driving needs and better refueling infrastructure provided in urban

areas. The interaction of the household-specific gasoline price with gallons/mile, which measures

the operating cost per mile of the vehicle, has a negative sign, suggesting that consumers have a

negative preference for fuel costs. The estimation results also suggest that consumers who have

more education and live in cities with more charging stations are more likely to adopt EVs.

We include four random coefficients, which represent unobserved consumer preferences for

fuel economy (gallons/mile), acceleration (horsepower/weight), light trucks, and AFVs. Based

on the standard normal distribution of the random taste variable vik, the coefficient βuk can be

interpreted as the standard deviation in the unobserved preference for the vehicle attribute

k. To reduce simulation noise and bias, following Train and Winston (2007), we use 150

Halton draws to approximate the distribution for the unobserved consumer taste v.17 All four

coefficients are statistically significant, indicating that consumers have heterogenous preferences

for those vehicle attributes conditional on the observed consumer characteristics. Those precisely

estimated random coefficient parameters help alleviate the well-known problem of independence

of irrelevant alternatives experienced in traditional logit models and play a critical role in defining

the substitution patterns.

To illustrate the importance of estimating the model parameters with the second-choice

data, we re-estimate the first-stage parameters (the observed and unobserved interaction terms)

without these data. Recall that we estimate the model parameters with five years of data where

for each household observation we observe a second choice. The unobserved heterogeneity is

identified from both changes in choice sets and vehicle attributes over time and the correlation

between first- and second-choice vehicle attributes. Therefore, removing the second-choice data

allows us to test the importance of including these data relative to exploiting panel variation

that is traditionally used to identify unobserved heterogeneity (Berry et al., 1995). Our model

also has household demographics interacted with vehicle characteristics, and we include these as

well to isolate the impact of using the second choice data. Petrin (2002) finds that including

household demographic by vehicle characteristics interactions is crucial for obtaining precise

estimates of the unobserved heterogeneity terms. Therefore, we keep them in the model for each

17Halton draws are standard normal draws from Halton sequences. Halton sequences (Halton, 1960) are designedto induce a negative correlation over observations. Since each subsequence fills in the gaps of the previoussubsequences, the Halton draws fill-in the unit interval more evenly and densely, and provide better coverageand are more efficient than random draws for simulation. See Section 9.3.3 of Train (2009) for a more detaileddiscussion of Halton draw. The demand results are similar when the number of Halton draws is increased to 200.

16

set of estimates.

The first-stage parameter estimates without the second-choice data are shown in panel (b)

of Table 3 under “(2) No 2nd Choice.” Overall, the signs and magnitudes of the observed

heterogeneity terms are consistent with the estimated parameters with the second-choice

data included. These terms also maintain a high level of statistical significance, suggesting

that incorporating the second-choice data is not necessary for precise estimation of observed

heterogeneity. This is intuitive because the observed heterogeneity terms are identified from

correlations between observed household demographics and vehicle characteristics, which are not

dependent on the second-choice data.

Comparing the estimates of the standard deviations of preference parameters reveals striking

differences between the two models. Without the second-choice data, all but one of the standard

deviations becomes insignificant, and the magnitudes are much different from the parameters

estimated with the second-choice data. This reveals that we are unable to obtain precise

measures of unobserved heterogeneity without including the second-choice data, even with panel

variation from five years of rapidly changing vehicle attributes and choice sets. To the best of

our knowledge, this is the first illustration of the value of exploiting repeated choice data to

identify unobserved heterogeneity parameters relative to the value of using panel variation to

identify these parameters. The lesson from this comparison is clear: the second-choice data

greatly improve the statistical precision of the unobserved heterogeneity parameters relative to

a model estimated using the standard panel variation approach.18

5.2 Elasticities and Substitution Patterns

The demand system implies sensible price elasticities. All of the implied own-price elasticities

are greater than one, ranging from -3.97 to -2.37 with the average being -2.67 and the standard

deviation being 0.21. The sales-weighted average elasticity among all 2,146 products in five

model years is -2.75. The magnitude of the own-price elasticities is close to the one obtained by

Durrmeyer and Samano (2017) and slightly smaller than those obtained by Berry et al. (1995),

Petrin (2002), Beresteanu and Li (2011), and Li (2012), since our demand estimation is based on

consumers who purchase new vehicles (excluding outside option). Our estimate is close to that

of Train and Winston (2007), who also estimate the demand focusing on new vehicle buyers and

find an average own-price elasticity of -2.32. Appendix Figure A.2 plots the own-price elasticities

18In Appendix Table A.8, we also provide the results of key counterfactual analysis based on the demandestimates without using the second choice data.

17

against price and demonstrates that more expensive models tend to have less elastic demand.

Our own-price elasticity of demand estimate for EVs is -2.76. This estimate is smaller in

magnitude than the estimate from Muehlegger and Rapson (2020) who report an own-price

elasticity estimate being -3.2 to -3.4. Our estimate differs from theirs for a few reasons. First, our

estimate applies to all new vehicle buyers, which contrasts with Muehlegger and Rapson (2020)

which is estimated from vehicles bought by low and middle income households. Conditioning on

buyers having income less than $100,000, our conditional own-price elasticity of demand for EVs

becomes -3.13, which is more comparable to theirs.19

Table 4 shows the cross-price elasticities for a selected group of models. One obvious pattern

is that the demand for less expensive models tends to be more price sensitive. More expensive

models such as Tesla Model S have lower own-price elasticities in magnitude. Compared with

other conventional gasoline vehicles, electric vehicles such as the Nissan LEAF and the Chevrolet

Volt have larger cross-price elasticities with hybrid vehicles such as Toyota Prius. Battery electric

vehicles such as the Nissan LEAF and the Tesla Model S do not have large cross-price elasticities

with plug-in hybrid vehicles such as the Chevrolet Volt. BEVs can only run on electricity and

many of them have limited range. PHEVs, on the other hand, rely on gasoline mode to boost

the range, since the electric range is only around 30-40 miles. These two different kinds of plug-

in vehicles are likely to attract consumers with different driving needs as consumers with more

frequent long-distance travels are more likely to adopt PHEVs. Therefore, it makes sense that

no strong substitution exists between PHEVs and BEVs, especially when there were only a few

models during the early deployment stage. Ford F-150, the only pickup truck in the selected

sample, does not have much substitution with the other small and mid-size sedans, and it has

almost zero substitution with EV models. The substitution pattern indicates that consumers

who purchase EVs generally favor mid-size sedans that are relatively fuel-efficient rather than

large vehicles.

Table 5 summarizes the elasticity estimates by fuel type. Across different fuel types, the

sale-weighted own-price elasticities are similar since all fuel types include vehicle models with a

large price range. Each cell in the matrix represents the average sales change of a vehicle model

in that fuel type from a price change of a vehicle model of other fuel types. For example, a

10 percent increase in the price of hybrid vehicle model will increase the sales of a BEV model

by 0.37 percent on average, and a 10 percent increase in the price of another BEV model will

19The income eligibility requirement in Muehlegger and Rapson (2020) is that a households must have anincome less than 400 percent of the federal poverty line. A household income of $100,000 is roughly 400 percentabove the 2014 poverty line for a family of four.

18

increase the sales of a BEV model by 0.13 percent. Both BEVs and PHEVs have a larger cross-

price elasticity with respect to hybrid vehicle models relative to the cross-price elasticity with

respect to gasoline, diesel, and flexible-fuel vehicle (FFV) models, suggesting that EV buyers

prefer vehicles with better fuel economy. Because of the large selection of model choices, gasoline

vehicles are a major substitute for vehicles of all fuel types. Since our data mostly cover the first

few years after the introduction of EVs, the within-segment substitution for BEVs and PHEVs

is relatively small, considering that we do not have enough between-segment variation to identify

a strong substitution between EV models.

6 Counterfactual Analysis

In this section, we conduct simulations to examine the counterfactual vehicle fleet where we

remove all EV models from the choice sets and where the EV subsidy were removed. The

magnitude of the resulting sales changes of the other fuel types could suggest what types of

vehicles were replaced by EVs. The estimated substitution patterns are then translated into

emissions reductions to assess the environmental benefits of EVs and EV subsidies.

6.1 The Environmental Benefits of EVs

The introduction of EVs could lead consumers who would originally choose gasoline or hybrid

vehicles to purchase EVs, and the substitution pattern critically determines the environmental

benefits of promoting EVs. To examine the substitution pattern of EVs with other fuel types, we

conduct a counterfactual exercise where all EVs are removed from the choice set. The resulting

changes in sales of other fuel types will reveal what types of vehicles EVs replace. Since we do not

allow consumers to choose an outside option, as the demand estimation is conditional on buying

a new vehicle, consumers who purchased EVs would switch to another non-EV model. In 2014,

109,449 EVs were sold in the US vehicle market. The simulation results suggest that 78.7 percent

of EVs replaced conventional gasoline vehicles, 12 percent of EVs replaced hybrid vehicles, 2.4

percent replaced diesel vehicles, and the remaining 6.9 percent replaced FFVs (see Table A.5).

The average fuel economy of the vehicles that were replaced by EVs is 28.9 mpg. This number

can be interpreted as the fuel economy level of the composite substitute of EVs, as defined in

Section 3. Among gasoline vehicles replaced by EVs, 74 percent of them have fuel economy

above 25 mpg. The vehicle models that were replaced by EVs most are: Honda Accord, Toyota

Prius, Toyota Camry, Honda Civic, Toyota Corolla, Nissan Altima, and Chevrolet Cruze. This

19

substitution pattern suggests that EVs mainly attracted consumers who were originally choosing

mid-size and fuel-efficient gasoline or hybrid vehicles, rather than gas-guzzlers such as large SUVs

or trucks. This substitution pattern can also be explained by the dominance of car models in

the EV category. All the EV models in our sample belong to the car segment and none are large

SUVs or pickup trucks.20 Appendix Figure A.1 shows that BEV and PHEV buyers are much

more likely to consider alternative vehicle choices within the car segment rather than the light

trucks segment that includes SUVs and pickups, compared with gasoline vehicle buyers.

To evaluate the environmental impact of the introduction of EVs, we evaluate the total

gasoline saved and CO2 emissions reductions from EVs by comparing the gasoline consumption

of the actual vehicle fleet with the counterfactual fleet without EVs. In order to compute the

emission impacts of EVs, we require an estimate of lifetime vehicle miles traveled (VMT) for every

vehicle. Following Linn (2020), we obtain vehicle odemeter data from 2017 National Household

Travel Survey (NHTS) and estimate the lifetime VMT for each vehicle segment and fuel type21.

Appdendix Section C provides a more detailed explanation of the procudure with the exact

estimates reported in Appendix Table A.6. Our simulation results show that the existence of

EVs helps save lifetime gasoline consumption of 0.40 billion gallons, resulting in a CO2 emissions

reduction up to 7.84 millions pounds, assuming a lifetime VMT for each vehicle segment and fuel

type. If we do not estimate the substitution pattern but assume each EV replaces a conventional

gasoline vehicle with fuel economy of 23 mpg and an average lifetime VMT22, the total lifetime

gasoline saved would become 0.65 billion gallons, with an emissions reduction of 12.8 billion

pounds of CO2. Simply assuming EVs replace a gasoline vehicle of an average mpg and VMT

would overestimate the environmental benefits of EVs by 39 percent. The overestimated portion

would be larger if EVs replace a greater number of fuel-efficient vehicles such as hybrid vehicles.

If EVs were removed from the market, consumers who value the EV technology will suffer

a welfare loss. We find that the removal of EVs from the choice set leads to a total consumer

welfare loss of $670.3 million in 2014.23 Panel (b) of Table A.5 summaries the impact of the

20By the end of 2018, 15 out of the 16 BEV models on the US market are cars and 24 out of the 29 PHEVmodels are cars.

21The VMT predictions show that EVs tend to have relatively low lifetime VMT, which is consistent with Davis(2019), who finds that EVs are driven less than their gasoline counterparts.

22According to our VMT prediction in Appendix Section C, the average lifetime VMT of gasoline vehicles is165,584

23The average welfare loss per household is estimated as the change in consumer surplus as shown by Smalland Rosen (1981). Total welfare loss is calculated as average consumer surplus loss multiplied by the market sizeof new vehicles.

20

removal of EVs on consumer surplus by income quintile. The results reveal a greater impact on

wealthier households, since they are more interested in the EV technology and thus benefit more

from the introduction of EVs.

6.2 Impacts of Income Tax Credits

The federal government has adopted several policies to support the EV industry including

providing federal income tax credits for EV purchases, R& D support for battery development,

and funding for expanding charging infrastructure. Congressional Budget Office (CBO)

estimates that the total budgetary cost for those policies will be about $7.5 billion through

2017. The tax credits for EV buyers account for about one-fourth of the budgetary cost and are

likely to have the greatest impact on vehicle sales. Under the tax credits policy, EVs purchased

in or after 2010 are eligible for a federal income tax credit up to $7,500. Most popular EV models

on the market are eligible for the full amount. The credit will expire once 200,000 qualified EVs

have been sold by each manufacturer. In 2014, the federal government spent $725.7 million on

income tax credits for EV buyers. To examine the effectiveness of the income tax credit policy

in terms of stimulating EV sales, we use our parameter estimates to simulate the counterfactual

sales of EVs that would arise in the absence of the tax credits to EV buyers in 2014. The

counterfactual sales could help us identify the percentage of “non-additional” EV sales and also

evaluate the environmental benefits of the policy. The resulting sales increase in gasoline and

hybrid vehicles could help us evaluate the environmental benefits of the “additional” EV sales.

The short-run benefits could be small if the additional sales simply come from people who were

considering buying other fuel-efficient vehicles.

The simulation results of the federal EV subsidy are summarized in Table 6. Our estimates

imply that removing the federal income tax credits reduces EV sales by 28.8 percent in 2014,

with BEVs experiencing a sales reduction of 32.6 percent and PHEV sales falling by 24.5 percent.

The results suggest that about 70 percent of the EV buyers would still purchase EVs without

income tax credits. Since the EV subsidy lowers the effective price of purchasing EVs, consumers

would enjoy a welfare increase due to the subsidy, especially for those who purchase EVs. Our

estimation results suggest that the EV subsidy program leads to a total increase in consumer

surplus of $165.2 million. The average increase of consumer surplus per household due to policy

may not be large, since most households do not purchase EVs. Panel (b) of Table 6 summarizes

the incidence of the federal EV subsidy. The distribution impacts imply that the EV subsidy

program is regressive, as higher-income households benefit more from the subsidy because they

21

are more likely to purchase EVs and thus claim the subsidy.

If there were no federal EV subsidy, 78.9 percent of the “additional” EV buyers would switch

to gasoline vehicles with an average fuel economy of 27.2 mpg, and 11.8 percent would switch to

hybrid vehicles with an average fuel economy of 45 mpg, with the remaining switching to diesel

and flex-fuel vehicles.24 Using the miles per gallon gasoline equivalent (MPGe) introduced by the

US Environmental Protection Agency (EPA), we can then translate those substitution patterns

to energy consumption reduction from the increased sales of EVs.25 By inducing consumers to

switch to more fuel-efficient EVs, the income tax credit policy leads to a reduction in lifetime

gasoline consumption of 0.1 billion gallons and CO2 emissions reduction of 2.02 billion pounds,

which is equivalent to reducing 14,309 gasoline vehicles of an average fuel economy of 23 mpg. If

we assume each EV replaced an conventional gasoline vehicle with a fuel economy of 23 mpg and

an average life VMT, the gasoline consumption saved would become 0.15 billion gallons and the

CO2 emissions would become 2.95 billion pounds, equivalent to removing 20,863 gasoline cars

from the road. Not taking account of the actual substitution pattern would overestimate the

environmental benefits of the EV subsidy by 32 percent (Figure 3).

Appendix Table A.7 summarizes the environmental benefits of EV income tax credits by

evaluating the external cost savings from emissions reduction of various pollutants. In 2014,

the EV subsidy results in total environmental benefits of $50.9 million from a more fuel-efficient

vehicle fleet, by taking account of the reduction of CO2, VOC, NOx, PM2.5, and SO2.

The environmental benefits and increase in consumer welfare are much lower than the total

spending of $725.7 million, since the majority of the subsidies are non-additional and the

additional portion mainly induces consumers who would purchase fuel-efficient vehicles anyway

to switch. The current subsidy policy offers equal tax credit amounts to all buyers of the same

electric model. Alternatively, more credits could be given to lower-income households, with no

tax credits given to the highest-income households. This policy design would mimic the policy

reform of California’s CVRP, which intends to direct the incentives toward households that are

likely to value the rebates the most. The subsidy could also target first-time buyers, who may

not have a good sense of vehicle fuel consumption but are more sensitive to upfront costs. Please

note that there are several caveats regarding our analysis of the environmental impacts of the EV

24In Appendix Section F, we examine substitution patterns between EVs and hybrid vehicles to assess whetherEV subsidies are “crowding out” hybrid sales. We find that EV subsidies have had a tiny impact on hybrid sales.We also find that the removal of the federal hybrid subsidy has had a much larger impact on reducing hybridsales.

25The MPGe metric was introduced in November 2010 by EPA. The ratings are based on EPA’s formula, inwhich 33.7 kilowatt-hour of electricity is equivalent to 1 gallon of gasoline.

22

subsidy. We will discuss how those issues potentially change our results in detail in Section 7.

6.3 Alternative Subsidy Designs

As with other energy subsidy programs, the cost-effectiveness of EV incentives could be

undermined if the subsidies are poorly-targeted and are mainly taken up by wealthier consumers

who would buy EVs without subsidies. To further investigate the “additionality” of the current

federal income tax credits for EVs and whether better targeting could improve the effectiveness of

the program in terms of boosting EV demand, we conduct two policy simulations to compare the

current uniform subsidy with alternative subsidy designs that incorporate an income-dependent

structure.

Through CVRP program, California has been providing state-level subsidies to EV buyers

since 2010. The standard rebate amounts are $2,500 for BEVs and $1,500 for PHEVs. On March

29, 2016, CVRP started to implement income eligibility requirements such that households with

income levels above certain thresholds are no longer eligible for the EV rebate, while lower-

income households can claim an additional rebate of $2,000 on top of the standard rate. The

income caps for high-income households are set at $150,000 for single filers and $300,000 for joint

filers. The households whose income levels are less than or equal to 300 percent of the federal

poverty level are categorized as the lower-income consumers. The motivation for switching from

a uniform subsidy to an income-dependent subsidy is to make EVs more accessible to a larger

number of drivers, especially those in lower-income households and communities that are more

affected by air pollution. Since higher-income households are less sensitive to prices and have

a stronger preference for newest technologies, they are more likely to adopt EVs without the

subsidy. Therefore, providing more generous subsidies to lower-income households that are more

price-sensitive could potentially reduce the policy cost of increasing EV sales.

The current federal-level EV subsidy is not designed to favor low-income households. To

investigate whether an income-dependent structure could be more effective in terms of inducing

additional EV sales, we conduct two counterfactual exercises that compare the current federal EV

subsidy with alternative subsidy policies that mimic the design of CVRP. We remove the subsidy

for households with income levels above the defined thresholds in both alternative policies, and

we provide lower-income households with an additional subsidy of $2,000 in alternative policy 1

and $4,000 in alternative policy 2. The income cap and low-income groups are defined the same

23

way as in the CVRP.26

The simulation results are summarized in Table 7. Under the current uniform subsidy, the

government spent $0.73 billion in 2014 in subsidizing EVs, and a total number of 109,850

EVs were sold in the market, leading to an average spending of $6,630 per EV. With the

removal of the subsidy to the high-income group and increased subsidy of $2,000 to low-income

households, EV sales would decrease by 701, a 0.6 percent decrease. However, the total subsidy

spending decreases from $0.73 billion to $0.64 billion, a 12.3 percent decrease. As a result, the

income-dependent subsidy reduces the average spending per EV from $6,630 to $5,870, which

is equivalent to a saving of $760 (11.4 percent) per EV. The alternative subsidy is more cost

efficient because it better targets low-income households that are more price sensitive. This

result is consistent with the predictions of our stylized model of additionality, in that a subsidy

targeting more price-sensitive buyers will result in more additional sales (equation 11).

In the second alternative policy design, an additional subsidy of $4,000 is provided to low-

income households, and high-income groups whose incomes are above the thresholds are still not

eligible for the subsidy. The total EV sales under alternative policy 2 increase by 1,232 (1.12

percent) compared with the existing policy. Even though an additional subsidy of $4,000 is given

to low-income households, the total spending is $0.66 billion, which is 9.6 percent lower than the

total spending of the current subsidy. The average spending per EV is $5,970 for alternative 2,

leading to a saving of $660 (10 percent) per subsidized EV. The savings are still mainly from the

removal of the subsidies given to households with higher income, which would purchase EVs in

the absence of the subsidy. By inducing additional EV sales, all the subsidy designs result in a

reduction of CO2 emissions by replacing vehicles that are less fuel-efficient with EVs.

The current subsidy policy reduces the total CO2 emissions from the new vehicle fleet by

0.92 metric tons, and alternative policies 1 and 2 achieve a total CO2 emissions reduction of 0.89

and 0.96, respectively. The average cost of CO2 reduction is then estimated to be $795, $725,

and $693, respectively. Both of the alternative policies are less costly than the existing policy

in terms of reducing emissions. Although alternative policy 2 has a higher average subsidy cost

per EV than alternative 1, it achieves a lower cost per CO2 reduction by inducing more sales

of BEVs, which are more fuel-efficient. By providing more generous subsidies to low-income

households, small and mid-size BEVs become more affordable to those consumers who prefer

26The income cap is $150,000 for single filers and $300,000 for joint filers. Low-income households are definedas those with income less than or equal to 300 percent of the 2018 federal poverty level. For households with oneto eight persons, the combined household income must be less than $36,420, $49,380. $62,340, $75,300, $88,260,$101,220, $114,180, and $127,140 respectively (https://cleanvehiclerebate.org/eng/income-eligibility).

24

smaller vehicles. The BEV models that experience the highest sales increases in alternative 2 are

Nissan LEAF, Smart ForTwo EV, and Fiat 500e. Beresteanu and Li (2011) estimate the cost of

CO2 reduction to be $177 per ton from the income tax credits for hybrid vehicles. Our estimates

of the cost of CO2 reduction of EV subsides are larger, since the federal subsidy for EVs is more

than twice the amount of the hybrid subsidy. Our estimates also suggest that subsidizing EVs

is a relatively costly way to achieve the emissions reduction goals.

Panel (b) of Table 7 compares the distributional impacts of the three subsidy designs. The

current uniform subsidy is regressive, since it benefits higher-income households more, as they are

more likely to purchase EVs and claim the subsidy. Both alternative policies are less regressive

than the existing policy, since they eliminate the subsidy for households whose income levels

are above the threshold. Alternative policy 2 benefits the bottom income group the most, as

it gives the most generous subsidy to low-income households. In summary, compared with the

current uniform subsidy, the income-dependent subsidy designs are more effective in stimulating

EV demand and reducing emissions, and they could also be better justified on distributional

grounds.

6.4 Alternative Data and Parameter Assumptions

Our findings of the effectiveness of the federal EV subsidy is based on the conditions of the

2014 new vehicles market. Therefore, much of the EV subsidy may not be additional as the

consumers who bought EVs during our sample period are mostly early adopters who favor the

new technology and are less price sensitive. But the recent years have witnessed a considerable

expansion of the EV market, including the introduction of more affordable EV models with

relatively long range such as Tesla Model 3. As the technology improves, the prices of EVs

will drop and the available EV models will increase. The subsidy is then expected to produce

more additional EV sales as it covers a larger portion of the upfront cost, and more average

households who are more price responsive would be attracted to purchase EVs. However, we

think evaluating the subsidy impact during the early years of EV industry is informative as most

subsidy polices are implemented during the early deployment stage of newest technologies when

the adoption cost is high and early adopters are less price sensitive. To subsidize hybrid vehicles,

the federal government provided income tax credits of up to $3,400 for hybrid vehicles starting

December 31, 2005. But all the hybrid vehicles purchased after December 31, 2010, were no

longer eligible for this credit. The federal EV tax credits will also phase out once a automaker

reaches the 200,000 unit sales threshold, and the subsidies have already started to phase out for

25

Tesla buyers beginning 2019. Therefore, designing more effective subsidy policies to promote a

new technology and speed up its diffusion process is thus critical and policy relevant.

To determine how the changes of the recent and future EV industry affect our results, we

simulate different market outcomes without the EV subsidy for various parameter and data

assumptions. Table 8 summaries the sales and environmental impacts of the EV subsidy for

the different hypothetical scenarios. Columns (1) and (2) set the own price elasticity to -2 and

-4, respectively, to isolate the impacts of price sensitivity of the market base on the subsidy

effects. The simulation results show that as more average households join the EV market and

consumers become more price responsive, the EV subsidy has a larger impact by inducing more

additional EV sales.27 Columns (4) to (6) keep the estimated demand parameters but vary the

EV prices to reflect the price decrease of EVs due to falling battery prices. Results reveal that

the “additionality” of the subsidy increases as the EV prices fall. If the EV prices were 50% lower

than the actual prices in 2014, the federal EV income tax credits could increase the EV sales by

57%. Therefore, the EV subsidy would be more effective in increasing EV sales when consumers

are more price sensitive and falling input prices makes EVs less expensive. However, the average

MPG of the vehicles replaced by the additional EVs is not largely impacted by the consumer

sensitivity or EV prices, as EVs mostly replace vehicles belonging to the car segment and with

relatively high fuel economy. As more electric SUV models are introduced in the vehicles market,

we may expect EVs to replace more light trucks that are less fuel efficient.

7 Discussion

The results from our analysis come with several caveats. First, the estimates of the environmental

benefits are relatively crude, as they do not incorporate spatial heterogeneity of the upstream

emissions from electricity generation. As Holland et al. (2016) show, great spatial heterogeneity

exists regarding the environmental benefits promoting EVs, and in some locations where

electricity generation relies much on fossil fuels, EVs should be taxed rather than being

subsidized. The focus of our study is to demonstrate that the substitution pattern is also an

important factor in determining the environmental benefits, which matters even if we focus on

a specific location where the grid fuel mix is fixed. Nevertheless, different markets might reflect

different substitution patterns, which leads to different environmental benefits of promoting

EVs. Therefore, incorporating spatial heterogeneity and estimating location-specific substitution

27This result is consistent with our stylized model of additionality (equation(11)).

26

patterns would help us determine the location-specific environmental benefits of EVs, but it

would require more detailed and representative location-specific sales and consumer survey data.

However, our analysis provides empirical evidence of EV substitution at the national level and will

provide guidance for the federal government to evaluate the effectiveness of federal EV subsidies.

The findings of the paper would be policy-relevant since most markets worldwide subsidize EVs

at the national level.

Second, when estimating the environmental benefits of replacing gasoline vehicles with EVs,

we assume that vehicle miles traveled do not change in response to the EV subsidy. In fact, when

consumers switch to fuel-efficient vehicles with a lower marginal cost of driving, they might drive

more, resulting in a rebound effect that undermines some environmental benefits of EVs. On

the other hand, when consumers switch to smaller and lower-performance vehicles, the marginal

benefits of driving per mile could be reduced. Thus, the rebound effect could be weakened

by shrinking vehicle size and the net response of miles could be zero or negative (Anderson

and Sallee, 2016; West et al., 2017). Moreover, because of the limited range of EVs and the

inconvenience of charging in some locations, it is less likely that consumers would increase miles

traveled once adopting EVs.28 What is required to make this assessment is a joint model of

VMT and vehicle portfolio choice, where households make simultaneous decisions on how many

vehicles to own, which vehicles to own, and how much to drive each vehicle.29

Third, our demand estimation is conditional on consumers choosing a new vehicle without

considering the outside option, which includes used car markets or public transportation. An

alternative, more general formulation would be to model the choice between purchasing a new or

used vehicle, or the even broader decision to purchase a vehicle or not make a vehicle purchase

at all. These alternative models would allow for a richer set of substitutions resulting from

changes in the characteristics of EVs. EVs entering the market could increase total new vehicle

sales, as some households may decide to buy an additional vehicle (while continuing to drive

their other vehicles). This might increase fleet-wide emissions assuming no changes in vehicle

miles traveled.30 Estimating this effect would require variation in the introduction of EVs that

28Recent empirical evidence suggests that EVs are generally driven less than a comparable gasoline vehicle(Davis, 2019).

29Prior structural demand models tend to include an assumption that VMT is decided separately for eachvehicle owned by a household (Bento et al., 2009; Jacobsen, 2013a). These models could be extended to modelthe joint VMT decision to allow for the VMT substitution effect that we mention.

30The direction of the bias from omitting the outside option is uncertain. In this study, we assume the subsidymakes consumers switch from another new vehicle to an EV. However, in reality, consumers might switch from aused vehicle (more polluting than a new vehicle) or public transportation (less polluting than a new vehicle) toan EV due to subsidy. Therefore, the subsidy benefits might be under- or over-estimated, which depends on the

27

is exogenous of macroeconomic trends, such as increasing income during our sample period. To

the best of our knowledge, this has not been addressed in the literature. Recent estimates

for the substitution between new and used vehicles suggests that used vehicles are a weak

substitute for new vehicles (Leard, 2020). Therefore we would expect few used vehicle buyers

substituting to purchasing a new EV in response to the EV subsidy. Moreover, considering that

EV adopters favor the newest technologies and have a disproportionately high leasing rate (44

percent, Appendix Table A.3) and most of the EV survey respondents report another new vehicle

as their second choice, we believe that it is reasonable to assume that they would still choose

a new vehicle even if the subsidy were to be removed. Thus, also given that EVs only take

up a small market share of total vehicle sales, ignoring the outside option is unlikely to have a

significant impact on our results.31

Fourth, in both the demand estimation and the counterfactual simulations, we assume a

full pass-through of the EV subsidy. If this assumption is violated, we would underestimate

the true costs of consumers in acquiring EVs and we would overestimate the effects of the EV

subsidy. Evidence from other existing studies of related programs suggest that full pass-through

is a reasonable assumption for alternative fuel vehicle subsidies. Muehlegger and Rapson (2020)

find a 100 percent pass-through estimate of the EV subsidy using transaction data in California.

Gulati et al. (2017) find the pass-through of hybrid vehicle subsidies are close to 80-90 percent,

if accounting for quality upgrading. Sallee (2011) estimates the incidence of tax incentives for

Toyota Prius and find both federal and state incentives were fully captured by consumers. One

explanation provided in his paper is that Toyota believed future demand for hybrids would be

diminished if they had charged market clearing prices for the Prius in early years, and the

automaker set the prices low because they wanted the hybrid vehicle technology to be viewed as

an affordable and mainstream technology. In our research setting, the early deployment years of

EVs share many of the same market features as the hybrid vehicle market. Since manufacturers

mostly need to fulfill a minimum requirement of EVs because of increasingly stringent CAFE and

Zero Emission Vehicle (ZEV) regulations,32 it is not surprising that they fully pass the subsidy

to consumers, especially during the initial rollout years of EVs, which our sample period covers.

exact substitution with the outside option.31Previous studies also find a small extensive-margin response to price changes in other similar subsidy

programs.Beresteanu and Li (2011) estimate the impact of income tax credits for hybrid vehicles and find thatthe increase in total vehicles sales due to subsidy is only 0.26% of total new vehicle sales and only 1.1% of thehybrid vehicle sales are additional vehicle purchases due to the subsidy.

32The Zero Emission Vehicle (ZEV) program is adopted in California and nine other states, which requireautomakers to produce a number of ZEVs and plug-in hybrids each year, based on the total number of cars sold.

28

Another common feature of alternative fuel vehicle subsidies is that they are directly provided to

consumers through the forms of tax rebates and income tax credits, which might have different

incidence implications than production subsidies to manufacturers (Sallee, 2011).

Fifth, one limitation of our model is that it reflects short run changes in consumer behavior but

does not capture broader long run adjustments. We model the change in vehicle sales as a result of

changes in choice sets and changes in vehicle characteristics. We do not model adjustments made

by manufacturers in response to these changes, and our model does not include a characterization

of the electric vehicle charging station network. Prior work has found that manufacturers respond

to changes in subsidies by adjusting vehicle prices, and the vehicle charging station network

increases in size in response to subsidies (Beresteanu and Li, 2011; Li et al., 2017). If EVs

were removed from the choice set, closer substitutes to EVs such as hybrid vehicles and other

fuel-efficient gasoline models would be priced higher due to less competition they face. Thus, in

the counterfactual scenario where EVs had not been introduced, more consumers would switch

to less-fuel-efficient gasoline models than what we predicted. Therefore, by assuming away the

price effects on substitutes, we would underestimate the effects of EVs since we do not quantify

the additional benefits from lowering the prices of other competitive fuel-efficient substitutes.

However, this effect may not be large given that other policy constraint such as CAFE may

depress the prices of fuel-efficient vehicle models (Jacobsen, 2013b), and that, EV sales and the

number of EV models were limited during our sample period. The response of the vehicle charging

station network is likely to make the EV subsidy look more cost effective through its feedback

effect on the demand for EVs. Furthermore, EV subsidies are one instrument to use to internalize

the charging station network externality, which prior work has found to be a relevant barrier for

EV adoption (Li et al., 2017; Zhou and Li, 2018; Springel, 2020). These incentives also could

appear more cost effective based on the social benefit from internalizing technological innovation

spillover externalities. While the literature does not have any direct evidence on the magnitude of

this effect, work has shown that this benefit could be large based on evidence from other related

technologies such as solar power (Linn and McConnell, 2019). Despite these drawbacks, our

work accurately reflects relevant substitution patterns between EVs and conventional gasoline

vehicles, which are a key input in a model that incorporates these broader adjustments.

Another possibly relevant feature absent from our model are interactions with other

overlapping regulations. Subsidizing electric vehicles raises the demand for relatively fuel efficient

vehicles. Given that federal CAFE and GHG standards for light-duty vehicles are likely binding

during our time period, the federal EV subsidy effectively relaxes the CAFE and GHG constraints

29

faced by vehicle manufacturers. As a result, manufacturers are able to sell more relatively fuel

inefficient vehicles. Therefore, the short run effect of the subsidy on average fleet-wide fuel

economy and GHG emissions could be small or even zero if the standards are strictly binding

for all manufacturers. This result, however, assumes that the CAFE and GHG standards are

set in isolation of setting the EV subsidy. In fact, policy makers could set more stringent CAFE

and GHG standards in response to a more generous federal subsidy, which would maintain the

emissions benefits from the EV subsidy. We leave analyzing these possible responses for future

work.

Finally, our simulations are of counterfactuals for the model year 2014. Several features of the

electric power sector and the EV market have changed since 2014 that likely have an impact on

the outcomes that we document. Here we discuss how our results could be different with using

more recent data. First, the electric power sector has become less GHG emissions intensive over

time. This increases the GHG emissions reductions from the replacement of gasoline vehicles

with EVs, as discussed in Holland et al. (2019). Therefore, the current replacement of gasoline

vehicles with EVs should yield greater emissions reductions than what we find in our simulation

results. Second, the Tesla Model 3 entered the new vehicles market in 2018 and has since been

the highest selling EV. This is partly due to the fact that the Tesla Model 3 has been marketed

as a relatively affordable EV. Given its lower purchase price, the federal subsidy is likely to have

had a larger impact on increasing sales of this vehicle, as more mainstream new vehicle buyers

become marginal to buying affordable options like the Model 3. We document this effect in our

simulations in Section 6.4 where we find reducing purchase prices of EVs increases the fraction

of additional EV buyers. This effect is likely to persist or grow over time as battery prices

are expected to continue falling, translating into lower EV prices. Third, substitution patterns

between EVs and non-EVs may be different based on the composition of the choices offered to

consumers. The new vehicle market now has more EV models for consumers to choose from,

including an SUV (the Tesla Model X). Our simulations are based on a time period where only

EV models that were available were sedans. The expansion of EVs into SUV and light truck

segments could alter the substitution patterns that we have documented. As EVs become more

affordable and enter more segments over time, we expect to see the replacement gasoline vehicle

reflect characteristics of a more “average” vehicle.

30

8 Conclusions

Promoting electric vehicles is considered an effective way to increase fleet fuel economy and

reduce emissions from on-road transportation. The environmental benefits of subsidizing EVs

critically hinge on the fuel efficiency of the substitute vehicles. Encouraging consumers who would

otherwise purchase another fuel-efficient vehicle to switch to EVs would not lead to significant

emissions reductions.

The paper provides a theoretical and empirical analysis on how substitution patterns between

vehicles of different fuel types affect the emissions impacts of electric vehicle policies. The styled

theoretical model shows that the emissions impacts crucially depend on own-price and cross-price

elasticities of demand, where the emissions of non-EVs with a larger cross-price elasticity have

a bigger impact on the emissions impacts of EVs. To characterize the price elasticities and the

substitution pattern, we estimate a flexible discrete choice model of new vehicle demand that

incorporate rich consumer heterogeneity. A key differentiating feature of our demand model is

that we identify random preference heterogeneity by using the second-choice information from

household survey data, which greatly improve the precision of estimated preference heterogeneity

and implied substitution patterns. Our simulation results suggest that 79 percent of EVs replace

gasoline vehicles with an average fuel economy of 27.2 mpg, and 12 percent of EVs replace hybrid

vehicles with an average fuel economy of 45 mpg. If we had simply assumed that each EV replaces

an average gasoline vehicle of 23 mpg, we would have overestimated the environmental benefits

of EVs by 39 percent.

Our estimates imply that in 2014, the federal income tax credit for EVs led to a 28.8%

increase in EV sales, the majority of which replaced vehicles that are relatively fuel-efficient.

The increased EV sales translate to $73.8 million of environmental benefits due to reduced

emissions of major air pollutants. The cost-effectiveness of the policy is hindered by the fact

that about 70 percent of consumers would purchase EVs in the absence of the subsidy, and the

subsidy mainly attracted consumers who would otherwise have purchased fuel-efficient gasoline

or hybrid vehicles. By comparing the current uniform subsidy with alternative policy designs that

limit eligibility and provide additional subsidies to low-income households, we find the income-

dependent subsidies could potentially increase the cost-effectiveness of the subsidy program and

are less regressive. Policies intended to promote EV technology and reduce emissions would

be more effective by better targeting marginal buyers and encouraging consumers who would

otherwise purchase gas-guzzlers such as large SUVs to adopt EVs.

31

Table 1: Household and Vehicle Summary Statistics

2010 2011 2012 2013 2014 All YearsVariables Mean S.D. Mean S.D. Mean S.D. Mean S.D. Mean S.D. Mean S.D.Household income (1,000$) 146.78 107.92 160.10 134.60 130.65 104.49 136.78 115.16 136.21 107.96 140.45 114.16Household size 2.62 1.21 2.67 1.20 2.71 1.23 2.66 1.21 2.65 1.21 2.66 1.21With a college degree 0.62 0.49 0.62 0.49 0.63 0.48 0.65 0.48 0.65 0.48 0.64 0.48Living in an urban area 0.65 0.48 0.64 0.48 0.66 0.47 0.66 0.47 0.68 0.47 0.66 0.47Average commuting time (min.) 25.78 5.81 25.57 5.84 25.48 5.78 25.72 5.75 25.52 5.67 25.60 5.76Average gasoline price ($) 2.75 0.16 3.42 0.42 3.65 0.27 3.67 0.22 3.57 0.23 3.48 0.40Average vehicle price (1,000$) 31.17 11.39 32.00 11.49 33.16 11.28 34.19 12.84 34.98 14.00 33.46 12.54Average mpg of the vehicle 23.38 5.57 28.51 18.65 36.80 27.87 39.78 27.05 38.06 25.56 34.81 24.50Purchasing a light truck 0.49 0.50 0.49 0.50 0.41 0.49 0.36 0.48 0.41 0.49 0.42 0.49Household observations 1,509 1,860 2,287 2,899 3,073 11,628

Horsepower/weight (hp/lb) 0.06 0.02 0.06 0.02 0.06 0.02 0.07 0.02 0.06 0.02 0.42 0.49Wheelbase*width (in2) 8,399 1,165 8,339 1,173 8,313 1,193 8,289 1,167 8,314 1,172 8,330 1,173% of ICE models 92.92 91.58 89.71 88.44 88.24 90.12% of hybrid models 7.08 7.67 8.85 8.62 8.28 8.11% of EV models 0.00 0.74 1.43 2.95 3.49 1.77Vehicle choice set size 424 404 418 441 459 2146

Notes: The household-level data represent a sample (11,628 observations) drawn from the MaritzCX household survey data. Data of vehicleattributes are obtained from Wards Automotive. Household income is converted to 2014 dollars using the BLS calculator. Household size is thenumber of individuals living in the respondent’s household. Gasoline prices are quarterly average national prices from the EIA. Average mpg of thevehicle represents the average miles per gallon of the vehicles bought by the household sample. Horsepower/weight measures a vehicle’s acceleration,and wheelbase*width measures a vehicle’s footprint. ICE (internal combustion engine) models include gasoline, diesel, and FFV models.

32

Table 2: Summary of second choices for EV buyers

Make Model Fuel type Top 1 second choice Top 2 second choiceHonda Accord Plug In Hybrid PHEV Tesla Model S Toyota PriusFord C-Max Energi PHEV Toyota Prius Chevrolet VoltFord Fusion Plug In Hybrid PHEV Chevrolet Volt Toyota Prius Plug InToyota Prius Plug-in PHEV Chevrolet Volt Nissan LEAFChevrolet Volt PHEV Toyota Prius Nissan LEAFFiat 500 Electric BEV Nissan LEAF Mini CooperMercedes-Benz B Class Electric BEV Nissan LEAF Ford Fusion HybridFord Focus Electric BEV Nissan LEAF Chevrolet VoltNissan LEAF BEV Chevrolet Volt Toyota PriusTesla Model S BEV Nissan LEAF Audi A7Toyota RAV4 EV BEV Nissan LEAF Tesla Model SChevrolet Spark Electric BEV Nissan LEAF Chevrolet VoltSmart fortwo electric BEV Nissan LEAF Chevrolet VoltMitsubishi i-MiEV BEV Nissan LEAF Ford Focus ElectricBMW i3 BEV Nissan LEAF Tesla Model S

Notes: The data summary is based on the sample of 2014 EV buyers from the MaritzCX household survey data.

The table summarizes the most popular alternative vehicle choices for the households that purchased different EV

models. Top 1 second choice indicates the most frequently reported alternative choices among the buyers of a

specific EV model. Top 2 second choice reports the second most reported alternative choices for each EV model.

33

Table 3: Demand Estimation Results

Panel (a): Mean Utility Parameters

(1) OLS (2) IVCoefficient S.E. Coefficient S.E.

constant -2.4560 0.1028 -2.4728 0.1113log(price) -1.3497 0.2572 -1.7375 0.8164horsepower/weight 1.9525 0.3211 2.2664 0.6870weight 1.3393 0.2505 1.2943 0.2606gallons/mile 0.1805 0.1283 0.1327 0.1529AFV dummy -3.6229 0.5823 -3.6632 0.5942EV dummy -2.9321 0.2652 -2.8913 0.2791pickup dummy 0.4192 0.2863 0.7285 0.6828model year 11 dummy 0.2743 0.1143 0.2763 0.1144model year 12 dummy 0.3127 0.1075 0.3321 0.1153model year 13 dummy 0.0559 0.1063 0.0687 0.1098model year 14 dummy 0.1741 0.0920 0.1768 0.0929

Panel (b): Heterogeneous Utility Parameters

(1) 2nd Choice (2) No 2nd choiceCoefficient S.E. Coefficient S.E.

Observed Heterogeneitylog(price)/income -9.0659 0.4839 -15.9223 0.8972family size*vehicle weight 0.0892 0.0207 0.2207 0.0315urban*pickups -0.6678 0.0561 -0.7232 0.0705urban*EV 0.2305 0.0482 0.3636 0.0583gasoline price*gallons/mile -0.3078 0.0234 -0.6481 0.0410education*EV 0.8309 0.0808 1.1437 0.0939stations*EV 0.6728 0.1022 0.7469 0.1190

Random coefficientsgallons/mile 1.9291 0.0565 1.5953 0.9018horsepower/weight 1.0865 0.0396 0.1136 0.2129light trucks 0.2823 0.0254 0.7164 0.1099AFVs 0.9493 0.0886 0.4692 0.3074

Average own-price elasticity -2.67Average own-price elasticity of EVs -2.76Average elasticity of EVs (income < $100,000) -3.13

Notes: The number of households is 11,628. The value of the simulated log-likelihood at convergence is -144,129.4

based on 150 Halton draws per household. The instrumental variables used to estimate the linear parameters are the

difference and squared difference in characteristics (fuel economy, horsepower, and weight) with other vehicles sold

by the same manufacturer and the squared difference in characteristics of vehicles sold by other manufacturers. For

EVs, the gallons/mile measurement is defined as the inverse of miles/gallon gasoline equivalent, which is a metric

defined by EPA to measure the energy efficiency of EVs. Specification (1) includes consumers’ second choices in

the likelihood function while specification (2) only incorporates consumers’ purchased choices in constructing the

likelihood.

34

Table 4: A sample of own- and cross-price elasticities

Nissan Chevrolet Honda Ford Nissan Tesla Chevrolet Toyota Honda Ford PriceProducts Sentra Cruze Civic Focus Leaf Model S Volt Prius Accord F-150 in 2014Nissan Sentra (gas) -3.01 0.06 0.06 0.06 0.05 0.03 0.04 0.05 0.05 0.02 13,351Chevrolet Cruze (gas) 0.07 -2.95 0.07 0.07 0.06 0.04 0.05 0.06 0.06 0.02 19,243Honda Civic (gas) 0.12 0.11 -2.91 0.11 0.10 0.07 0.08 0.09 0.09 0.03 20,106Ford Focus (gas) 0.07 0.06 0.06 -2.97 0.05 0.04 0.05 0.05 0.05 0.02 20,026Nissan LEAF (BEV) 0.01 0.01 0.01 0.01 -2.83 0.02 0.03 0.03 0.01 0.00 29,799Tesla Model S (BEV) 0.00 0.00 0.00 0.00 0.01 -2.37 0.01 0.01 0.00 0.00 74,935Chevrolet Volt (PHEV) 0.01 0.01 0.01 0.00 0.02 0.01 -2.66 0.01 0.00 0.00 35,203Toyota Prius (HEV) 0.04 0.04 0.03 0.03 0.10 0.07 0.09 -2.68 0.03 0.01 24,027Honda Accord (HEV) 0.12 0.12 0.12 0.12 0.10 0.09 0.10 0.10 -2.67 0.04 24,436Ford F-150 (gas) 0.02 0.02 0.02 0.02 0.02 0.02 0.02 0.02 0.02 -2.53 29,806

Notes: The table reports a sample of own- and cross-price elasticities which are calculated based on the parameter estimates reported in Table 3 with

the IV specification of the mean utility parameters and second choice specification of the heterogeneous utility parameters. The elasticity estimates are

calculated with the same individual weights and Halton draws used in the demand estimation. More details are provided in Appendix C. The last column

in the table gives the average transaction price in 2014 for those selected vehicle models. The sales-weighted average elasticity among all 2,146 products

in five model years is -2.75.

35

Table 5: Own and Cross-Price Elasticity of Demand Estimates by Fuel Type

BEV PHEV Hybrid Gasoline Diesel FFV

BEV 0.013 0.018 0.015 0.004 0.003 0.002PHEV 0.012 0.011 0.009 0.002 0.002 0.002Hybrid 0.037 0.036 0.029 0.009 0.007 0.006Gasoline 0.028 0.027 0.028 0.029 0.026 0.027Diesel 0.003 0.004 0.004 0.006 0.007 0.007FFV 0.012 0.012 0.013 0.021 0.024 0.024

Own Price Elasticity -2.751 -2.649 -2.705 -2.761 -2.606 -2.680

Notes: The table summarizes the sales-weighted average own- and cross-price elasticity

estimates by fuel type. The elasticity estimates are based on the parameter estimates

reported in Table 3. BEV stands for battery electric vehicle, which represents vehicles that

operate only with electricity, including a Tesla Model S. PHEV stands for plug-in hybrid

vehicle, which represents vehicles that are able to operate with either electricity or gasoline,

including a Honda Accord plug-in hybrid. FFV stands for flex-fuel vehicle, which represents

vehicles that are able to operate on E85 fuel. On average, a 1 percent increase in a BEV

model will increase the sales of other BEV models by 0.013 percent.

36

Table 6: Sales and Incidence Impacts of Removing EV Subsidies

Panel (a): Sales Impact

Fuel types Sales change Percentage changeEV -31,501 -28.8%BEV -18,861 -32.6%PHEV -12,640 -24.5%Other fuel types Sales change Percentage Change Percentage of EV sales reduction Average MPGGasoline 24,867 0.23% 78.9% 27.2Hybrid 3,728 0.92% 11.8% 45.0Diesel 741 0.17% 2.4% 27.5FFV 2,165 0.15% 6.9% 22.0All non-EVs 31,501 0.24% 100% 28.9

Panel (b): Private Consumer Surplus Impact

Income quintile(1) (2) (3) (4) (5)

Average consumer surplus loss per household ($) -9.38 -10.81 -12.30 -13.93 -16.75

Total consumer surplus loss (million $) -165.2

Notes: The table summarizes the market and consumer welfare impact of removing the federal-

level income tax credit for EVs in the year of 2014. In panel (a), the percentage of EV sales

reduction represents the percentage of the original EV purchasers that switch to a specific non-

EV fuel type if the federal subsidy were removed. Average mpg represents the average miles per

gallon of the vehicles that experience sales increase due to the removal of federal EV subsidy. In

panel (b), the average consumer surplus loss per household is calculated as the change in consumer

surplus as shown by Small and Rosen (1981) for the year 2014. Total consumer surplus loss is

calculated as average consumer surplus loss multiplied by the market size of new vehicles.

37

Table 7: Comparison of Current Subsidy with Alternative Designs

Panel (a): Sales and Environmental Impact

Current subsidy Alternative 1 Alternative 2Total spending (billion $) 0.73 0.64 0.66Total EV sales 109,853 109,152 111,085Total BEV sales 57,791 57,554 58,900Total PHEV sales 52,062 51,598 52,185Average spending per EV ($) 6,630 5,870 5,970Total CO2 reduction (mil. metric tons) 0.92 0.89 0.96Cost of CO2 reduction ($/metric ton) 795 725 693

Panel (b): Private Consumer Surplus Impact

Welfare change per household ($)Income Quintile Current subsidy Alternative subsidy 1 Alternative subsidy 21 9.38 11.73 14.732 10.81 11.07 11.403 12.30 12.30 12.304 13.93 13.29 13.295 16.75 8.99 8.99Consumer surplus change 165.2 million 154.4 million 163.9 million

Notes: The current subsidy policy provides uniform tax credits to all EV buyers. Both alternatives 1

and 2 set an income cap such that the highest income group is not eligible to claim the subsidy. Further,

Alternatives 1 and 2 provide an additional $2,000 and $4,000, respectively, to lower-income households,

compared with the current policy. In panel (b), the income groups are defined the same way as for the

income eligibility implemented in the CVRP in California. The consumer surlpus change per household

is calculated as the change in consumer surplus as shown by Small and Rosen (1981) for the year 2014.

38

Table 8: Subsidy Impacts with Alternative Data and Parameter Assumptions

(1) (2) (3) (4) (5) (6)actual market elasticity=-2 elasticity=-4 10% lower price 25% lower price 50% lower price

EV sales increase 28.8% 22.0% 42.4% 27.4% 34.6% 57.0%BEV sales increase 32.6% 24.5% 50.3% 31.2% 39.4% 65.1%PHEV sales increase 24.5% 19.2% 33.5% 23.1% 29.1% 47.8%Average MPG replaced 28.9 28.8 29.3 28.9 28.9 28.8Gasoline saved (billion gal.) 0.10 0.09 0.19 0.12 0.16 0.26CO2 saved (billion lb.) 2.02 1.94 3.66 2.41 3.04 5.05Equivalent gasoline cars reduced 14,309 13,720 25,871 17,014 21,524 35,729

Notes: The table summarizes the market and environmental impacts of removing the federal-level income tax credit for EVs in the year of 2014 for alternative

parameter and data assumptions. Sales increases are the increases of EV sales due to subsidy. Average MPG replaced is the average MPG of the vehicles

replaced by the additional EVs. Gasoline and CO2 savings are lifetime reductions achieved by the EV subsidy. Column (1) is based on the 2014 market

condition and the actual parameters estimated, as summarized in Table 6 and Figure 3. Column (2) and column (3) reset the own-price elasticity as -2 and

-4, respectively, while holding the market conditions as in 2014. Columns (4)-(6) keep the estimated demand parameters while lowering all the EV prices by

10%, 25%, and 50%, respectively.

39

Figure 1: Consumer Second Choices by Fuel Type

(a) Gasoline vehicle buyers (b) Hybrid vehicle buyers

(c) PHEV buyers (d) BEV buyers

Notes: The figure plots the frequency of alternative vehicle choices by fuel type for different groups of consumers based on the survey responses of

the 11,628 households in the sample. The number of observations for the buyers of gasoline vehicles, hybrid vehicles, PHEVs and BEVs is 9,295,

315, 1,246, and 772, respectively.

40

Figure 2: Sales Impacts of Removing EVs

Notes: The figure plots the sales impacts of removing all EVs from the choiceset in MY 2014 on other vehicles of different fuel types. The percentage changesrepresent the percentage of the total sales increase from non-EV models that eachfuel type contributes to. The average miles per gallon of the replaced vehicles andthe impacts of removing EVs on consumer surplus are summarized in AppendixTable A.5.

41

Figure 3: Environmental Benefits and Substitution

Notes: The figure reports the estimated lifetime emissions reduction achieved by the federal-level income tax creditfor EVs in 2014, by comparing the observed fleet with the simulated fleet when the federal subsidy is removed.Actual benefits are based on actual estimates reported in Table 6 and counterfactual benefits are based on thescenario that EVs only replace a gasoline vehicle with the fuel economy of 23 mpg (2014 average level). Theestimation of the energy reduction from the increased EVs due to subsidy is based on the miles per gallon gasolineequivalent (MPGe) provided by EPA. Equivalent reduction of gasoline cars represents the equivalent number ofgasoline cars with a fuel economy of 23 mpg that can be reduced by the increased EVs due to subsidy, in terms ofemissions.

42

References

Allcott, Hunt, Christopher Knittel, and Dmitry Taubinsky, “Tagging and Targeting of

Energy Efficiency Subsidies,” American Economic Review, Papers and Proceedings, 2015, 105

(5), 187–191.

Anderson, Soren and James Sallee, “Designing Policies to Make Cars Greener: A Review

of the Literature,” Annual Review of Resource Economics, 2016, 8, 157–80.

Archsmith, James, Alissa Kendall, and David Rapson, “From Cradle to Junkyard:

Assessing the Life Cycle Greenhouse Gas Benefits of Electric Vehicles,” Research in

Transportation Economics, 2015, 52, 72–90.

Bento, A., L. Goulder, M. Jacobsen, and R. von Haefen, “Distributional and efficiency

impacts of increased U.S. gasoline taxes,” American Economic Review, 2009, 99 (3), 667–699.

Bento, Antonio, Ravi Kanbur, and Benjamin Leard, “Designing efficient markets for

carbon offsets with ditributional constraints,” Journal of Environmental Economics and

Management, 2015, 70, 51–71.

Beresteanu, Arie and Shanjun Li, “Gasoline Prices, Government Support, and the Demand

for Hybrid Vehicles,” International Economic Review, 2011, 52 (1), 161–182.

Beresteanu, Ariel and Federico Zincenko, “Efficiency Gains in Rank-ordered Multinomial

Logit Models,” Oxford Bulletin of Economics and Statistics, 2018, 80 (1), 122–134.

Berry, Steven, James Levinsohn, and Ariel Pakes, “Automobile Prices in Market

Equilibrium,” Econometrica, 1995, 63 (4), 841–890.

, , and , “Differentiated Products Demand Systems from a Combination of Micro and

Macro Data: The New Vehicle Market,” Journal of Political Economy, 2004, 112 (1), 68–104.

Boomhower, Judson and Lucas W. Davis, “A Credible Approach for Measuring

Inframarginal Participation in Energy Efficiency Programs,” Journal of Public Economics,

2014, 113, 67–79.

Chandra, Ambarish, Sumeet Gulati, and Milind Kandlikar, “Green drivers of free riders?

An analysis of tax rebates for hybrid vehicles,” Journal of Environmental Economics and

Management, 2010, 60, 78–93.

Congressional Budget Office (CBO), “Effects of Federal Tax Credits for the Purchase of

43

Electric Vehicles,” 2012.

Davis, Lucas, “How Much are Electric Vehicles Driven?,” Applied Economic Letters, 2019, 26

(18), 1497–1502.

Diamond, David, “The Impact of Government Incentives for Hybrid-electric Vehicles: Evidence

from US States,” Energy Policy, 2009, 37 (3), 972–983.

Durrmeyer, Isis and Mario Samano, “To Rebate or Not to Rebate: Fuel Economy Standards

vs. Feebates,” 2017. Forthcoming in Economic Journal.

EPA, “Average Annual Emissions and Fuel Consumption for Gasoline-Fueled Passenger Cars

and Light Trucks,” 2008.

Fowlie, Meredith, Michael Greenstore, and Catherine Wolfram, “Are the Non-Monetary

Costs of Energy Efficiency Investments Large? Understanding Low Take-up pf a Free Energy

Efficiency Program,” American Economic Review, 2015, 105 (5), 201–04.

Gallagher, K. and E. Muehlegger, “Giving Green to get Green? The Effect of Incentives

and Ideology on Hybrid Vehicle Adoption,” 2007. Working Paper.

Goolsbee, A. and A. Petrin, “The Consumer Gains from Direct Broadcast Satellites and the

Competition with Cable TV,” Econometrica, 2004, 72 (2), 351–381.

Gulati, Sumeet, Carol McAusland, and James Sallee, “Tax Incidence with Endogenous

Quality and Costly Bargining: Theory and Evidence from Hybrid Vehicle Sales,” Journal of

Public Economics, 2017, (155), 93–107.

Halton, J.H., “On the Efficiency of Evaluating Certain Quasi-random Sequences of Points in

Evaluating Multi-dimensional Integrals,” Numerische Mathematik, 1960, 2, 84–90.

Heutel, Garth and Erich Muehlegger, “Consumer Learning and Hybrid Vehicle Adoption,”

Environmental and Resource Economics, 2015, 62 (1), 125–161.

Holland, Stephen, Erin Mansur, Nicholas Muller, and Andrew Yates, “Are There

Environmental Benefits from Driving Electric Vehicles? The Importance of Local Factors,”

American Economic Review, 2016, 106 (12), 3700–3729.

, , , and , “Distributional Effects of Air Pollution from Electric Vehicle Adoption,”

Journal of the Association of Environmental and Resource Economists, 2019, 6 (S1), 65–94.

Huse, Cristian, “The Market Impact and the Cost of Environmental Policy: Evidence from

44

the Swedish Green Car Rebate,” The Economic Journal, 2014, 124, F393–F419.

Ito, Koichiro, “Asymmetric Incentives in Subsidies: Evidence from a Large-Scale Electricity

Rebate Program,” American Economic Journal: Economic Policy, 2015, 7 (3), 209–37.

Jacobsen, Mark, “Fuel Economy and Safety: The Influences of Vehicle Class and Driver

Behavior,” American Economic Journal: Applied Economics, 2013, 5 (3), 1–26.

Jacobsen, Mark R., “Evaluating U.S. Fuel Economy Standards in a Model with Producer

and Household Heterogeneity,” American Economic Journal: Economic Policy, 2013, 5 (2),

148–187.

Kahn, Matthew, “Do greens drive Hummers or hybrids? Environmental ideology as a

determinant of consumer choice,” Journal of Environmental Economics and Managment, 2007,

54 (2), 129–145.

and Ryan Vaughn, “Green Market Geography: The Spatial Clustering of Hybrid Vehicles

and LEED Registered Buildings,” The B.E. Journal of Economic Analysis and Policy, 2009,

9 (2), 1935–1682.

Langer, Ashley, “(Dis)Incentives for Demographic Price Discrimination in the New Vehicle

Market,” 2012. Working Paper.

and Derek Lemoine, “Designing Dynamic Subsidies to Spur Adoption of New

Technologies,” 2018. working paper.

Leard, Benjamin, “Estimating Consumer Substitution Between New and Used Passenger

Vehicles,” Resources for the Future Working Paper, 2020.

Li, Jing, “Compatibility and Investment in the U.S. Electric Vehicle Market,” 2016.

Li, Shanjun, “Traffic Safety and Vehicle Choice: Quantifying the Effects of the ‘Arms Race’ on

American Roads,” Journal of Applied Econometrics, 2012, 27, 34–62.

, Lang Tong, Jianwei Xing, and Yiyi Zhou, “The Market for Electric Vehicles: Indirect

Network Effects and Policy Impacts,” Journal of the Association of Environmental and

Resources Economists, 2017, 4 (1), 89–133.

Linn, Joshua, “Reducing the Costs of Federal Fuel Economy and Greenhouse Gas Standards

by Accurately Estimating Lifetime Vehicle Miles Traveled.,” 2020. Resources for the Future

Issue Brief 20-10.

45

and Virginia McConnell, “The Role of State Policies under Federal Light-Duty Vehicle

Greenhouse Gas Emissions Standards,” Energy Policy, 2019, 126, 507–517.

Manski, Charles F. and Setven R. Lerman, “The Estimation of Choice Probabilities from

Choice Based Samples,” Econometrica, 1977, 45 (8), 1977–1988.

Muehlegger, Erich and David S. Rapson, “Subsidizing Low- and Middle-Income Adoption

of Electric Vehicles: Quasi-Experimental Evidence from California,” 2020. Working Paper.

Petrin, Amil, “Quantifying the benefit of new products: the case of minivan,” Journal of

Political Economy, 2002, 110 (4), 705–729.

Sallee, James, “The Surprising Incidence of Tax Credits for the Toyota Prius,” American

Economics Journal: Economics Policy, 2011, 3, 189–219.

Sexton, Steven and Alison Sexton, “Conspicuous Conservation: the Prius Halo and

Willingness to Pay for Environmental Bona Fides,” Journal of Environmental Economics and

Management, 2014, 67, 303–317.

Small, K. and H. Rosen, “Applied Welfare Economics with Discrete Choice Models,”

Econometrica, 1981, 49 (1), 59–80.

Springel, Katalin, “Network Externality and Subsidy Structure in Two-sided Market:

Evidence from Electric Vechile Incentives,” 2020.

Train, Kenneth, Discrete Choice Models with Simulation, Cambridge University Press, 2009.

Train, Kenneth E. and Clifford Winston, “Vehicle Choice Behavior and the Declining

Market Share of U.S. Automakers,” International Economic Review, 2007, 48, 1469–1496.

West, Jeremy, Mark Hoekstra, Jonathan Meer, and Steve Puller, “Vehicle Miles (Not)

Traveled: Fuel Economy Requirements, Vehicle Characteristics, and Household Driving,”

Journal of Public Economics, 2017, 145, 65–81.

Whitefoot, Kate, Meredith Fowlie, and Steven J. Skerlos, “Compliance by Design:

Industry Response to Energy Efficiency Standards,” 2013. Working Paper.

Zhou, Yiyi and Shanjun Li, “Dynamics of Technology Adoption and Critical Mass in the

Market for Electric Vehicles,” Journal of Industrial Economics, 2018, 2, 423–480.

46

Appendices

A Industry and Policy Background

There are currently two types of EVs for sale in the United States: battery electric vehicles

(BEVs) which run exclusively on high-capacity batteries (e.g., Nissan LEAF), and plug-in hybrid

vehicles (PHEVs) which use batteries to power an electric motor and use another fuel (gasoline)

to power a combustion engine (e.g., Chevrolet Volt). The deployment of both types of EVs

currently faces significant financial barriers: EVs are substantially more expensive than their

conventional gasoline vehicle counterparts.33 A key reason behind the cost differential is the cost

of the battery. Battery market analysts predict that as battery technology improves, the cost

should come down.

Governments have recently provided generous monetary and non-monetary incentives for

EVs.34 The US federal government provides income tax credits for new qualified EVs in the

range of $2,500 and $7,500 based on each vehicle’s battery capacity and the gross vehicle weight

rating. Several states add state-level incentives to further promote EV adoption. For example,

through the California Clean Vehicle Rebate Project (CVRP), California residents can receive a

rebate of $ 2,500 for purchasing or leasing a BEV and $1,500 for a PHEV, and the rebate amount

increases to $4,500 and $3,500, respectively, for lower-income consumers.

There are at least two challenges that could undermine the effectiveness of the subsidy policy.

First, the uniform subsidy to EV buyers may not always result in additional EV sales in the sense

that many of the buyers who claim the subsidy may still purchase EVs even if there were no

subsidy policy. Since early adopters of EVs are those who favor the newest technology, have the

strongest environmental awareness, and usually have higher income, it is more likely that the

effect of a uniform subsidy policy, such as the current federal EV income tax credit, on boosting

additional EV sales is limited.35

33For example, the manufacturer’s suggested retail price (MSRP) for the 2014 Honda Accord Hybrid is $29,945,while the 2014 Honda Accord Plug-In Hybrid is listed at $40,570, which is over a $10,000 difference.

34Several cities in China such as Beijing implement a license restriction policy for the registration of new vehiclesand some PEV models are exempt from this restriction.

35The CVRP used to offer incentives of $1,500 for PHEVs and $2,500 for BEVs, but the majority of the rebateswent to high-income households. To direct the rebates toward households that value the rebates most, CVRPhas been redesigned such that lower-income households will be able to claim a larger rebate. Households withincome less than 300 percent of the federal poverty level will be able to get $3,000 for PHEVs and $4,000 forBEVs, and households with gross annual income above certain thresholds -$250,000 for single filers, $340,000 forhead-of-household filers, and $500,000 for joint filers- are no longer eligible for the rebates.

47

The second challenge has to do with the type of vehicles that are replaced by electric vehicles.

A potential efficiency loss could arise if the subsidy does not induce people to switch from a gas

guzzler to an EV, but from another fuel-efficient gasoline vehicle, or another hybrid vehicle to

an EV, making little net gain of environmental benefits. Holland et al. (2016) evaluate the

heterogenous environmental benefits of EVs by comparing the externalities of EVs with their

gasoline counterparts. However, the relative environmental benefits would be smaller if a higher

fuel-efficient vehicle such as a hybrid vehicle is compared. At the national average fuel mix,

BEVs and PHEVs do not have an advantage over hybrid vehicles in emissions reduction, and

PHEVs even generate more emissions than hybrid vehicles (Appendix Table A.2). With the

expiration of the tax credits for hybrid vehicles, the income tax credits for EVs are likely to

encourage consumers who would otherwise purchase hybrid vehicles to purchase EVs. Appendix

Table A.1 shows that as the market share of EVs increases in most recent years, the market

share of hybrids starts to decline. Chandra et al. (2010) find that the rebate programs in Canada

primarily subsidize people who would have bought hybrid vehicles or fuel-efficient cars in any case

and may not be the most effective way to encourage people to switch away from fuel-inefficient

vehicles like large SUVs or luxury sport passenger cars, at least in the short or medium run.

One of the justifications for EV subsidies is to reduce the emissions from the transportation

sector by replacing fuel-inefficient vehicles with EVs. When life-cycle emissions are accounted

for, however, substantial heterogeneity in environmental benefits could exist. For example,

EVs may not have an advantage over conventional vehicles in locations where the electricity is

generated through fossil fuels. Thus, even if the EV subsidy results in additional EV purchases,

the reduction of overall emissions would be limited. By incorporating spatial heterogeneity

of damages and pollution export across jurisdictions, Holland et al. (2016) find considerable

heterogeneity in environmental benefits of EV adoption depending on the location and argue

for regionally differentiated EV policy. They find that the environmental benefits of EVs are

the largest in California because of large damages from gasoline vehicles and a relatively clean

electric grid, but the benefits are negative in places such as North Dakota where the conditions

are reversed.

B Additional Literature Review

Our analysis contributes to the literature on the diffusion of vehicles with advanced fuel

technologies (e.g., hybrid vehicles) and alternative fuels (e.g., FFVs). Kahn (2007), Kahn and

Vaughn (2009), and Sexton and Sexton (2014) examine the role of consumer environmental

48

awareness and signaling in the market for conventional hybrid vehicles. Heutel and Muehlegger

(2015) study the effect of consumer learning in hybrid vehicle adoption, focusing on different

diffusion paths of Honda Insight and Toyota Prius. Several recent studies have examined the

impacts of government programs at both the federal and state levels in promoting the adoption of

hybrid vehicles, including Diamond (2009), Beresteanu and Li (2011), Gallagher and Muehlegger

(2007), and Sallee (2011). These studies consistently find that better environmental awareness,

higher gasoline prices, and more generous incentives are associated with higher adoption of

“green” vehicles. We add to this literature by estimating substitution patterns among gasoline-

powered vehicles, hybrids, and EVs. To the best of our knowledge, our results are the first

to provide cross-price elasticities among these vehicle types, which will allow researchers and

policy-makers to determine how subsidy designs alter the sales mix of different power types.

Our analysis relates to the literature that studies the cost-effectiveness of energy subsidy

programs. Allcott et al. (2015) find that some energy efficiency subsidies are poorly-targeted

and are primarily taken up by consumers who are wealthier and more informed about energy

costs. They conclude that restricting subsidy eligibility could increase the welfare gains from

those subsidies. Boomhower and Davis (2014) find that half of all participants would have

adopted the energy-efficient technology even with no subsidy. Ito (2015) shows that most of

the treatment effects of incentives come from consumers who are closer to the target level of

consumption, and the treatment effect is not significantly different from zero for consumers

who are far from the target level. Fowlie et al. (2015) find evidence that high non-monetary

costs contribute to the low participation of energy efficiency investment for households and

there are demographic differences between households that chose to participate on their own

and those that were encouraged to participate in the program by encouragement intervention.

Langer and Lemoine (2018) investigate the efficient subsidy schedule for durable goods in a

dynamic setting and show that an efficient subsidy often increases over time and that consumers’

rational expectations of future subsidies and technological progress could substantially increase

public spending. By empirically estimating a national vehicle demand model in a static setting

and obtaining consumer preference parameters, we are able to run counterfactual simulations

to examine the cost-effectiveness of the current EV subsidy design and compare it with other

possible designs.

49

C Prediction of Lifetime Vehicle Miles Traveled

To compute the emissions effects of EVs and the EV federal tax credit, we require an estimate

of lifetime vehicle miles traveled (VMT) for every vehicle in our 2014 sample. We adopt the

methodology in Linn (2020) to assign lifetime VMT. For this methodology, we obtained vehicle

odometer data from the 2017 National Household Travel Survey (NHTS) and fit odometer models

based on vehicle age, segment, and fuel type. For each vehicle segment (s) - car, van, SUV, and

pick up truck - we estimated the following odometer model:

oj = fs(aj) +Xjβ + εij

where j indexes the vehicle, fs(aj) is a 4th order polynomial function of the vehicle’s age aj,

Xj is a vector vehicle fuel typeswith coefficient vector β , and εij is an error term. We follow

Linn (2020) and estimate lifetime VMT using the predicted odometer reading at age 25.

The estimates of lifetime VMT appear in Appenxidix Table A.6. The estimates conform with

expectations and are broadly consistent with results from Linn (2020). Pickup trucks and SUVs

tend to have higher lifetime VMT than cars. EVs tend to have relatively low lifetime VMT,

which is consistent with Davis (2019).

D Defining the Simulated Log-Likelihood and Gradient

The utility of household i purchasing vehicle model j is defined as:

uij =K∑k=1

xjkβk − α1lnPj + ξj︸ ︷︷ ︸δj

+α2lnPjYi

+∑kr

xjkzirβokr +

∑k

xjkvikβuk︸ ︷︷ ︸

µij

+εij,

where δj is the mean utility of vehicle model j, which is constant across consumers, and

µij is the individual-specific utility component related to observed and unobserved consumer

demographics. Let θ = {βokr, βuk} be the non-linear parameters that are estimated from the first

stage. The last component, εij, is the idiosyncratic preference of household i for vehicle model j

and is assumed to have an i.i.d. type 1 extreme value distribution. The probability of household

i purchasing j and considering h as the second choice is:

50

Pijh =

∫exp[δj(θ) + µij(θ)]

1 +∑

g exp[δg(θ) + µig(θ)]· exp[δh(θ) + µih(θ)]∑

g 6=j exp[δg(θ) + µig(θ)]f(v)dv.

Note that the choice set of the second choice excludes the purchased product j. Since

the random taste vik is not observed, the probability above is calculated by integrating over

the distribution of the unobserved taste v. This high-dimensional integration is estimated via

simulation where R denotes the number of simulation draws and the subscript r denotes a specific

random draw r:

Pijh =1

R

R∑1

exp[δj(θ) + µijr(θ)]

1 +∑

g exp[δg(θ) + µigr(θ)]· exp[δh(θ) + µihr(θ)]∑

g 6=j exp[δg(θ) + µigr(θ)]

=1

R

R∑1

AjrBr

· AhrB−jr

=1

R

R∑1

Pijr · P−jihr ,

where B−jr = Br − exp[δj(θ) + µijr(θ)], which denotes the choice set for the second choice

conditional on choosing j as the first good, and P−jihr is the probability of choosing h from a choice

set that excludes j conditional on a set of random draws r.

Let lnRi = lnPijh = lnPij + lnP−jih = ln( 1R

∑R1 Pijr) + ln( 1

R

∑R1 P

−jihr), the individual log-

likelihood of household i choosing the observed purchased model j and considering the observed

second choice h. The log-likelihood function of the entire sample for a single market is therefore:

lnL =N∑i=1

lnRi.

The derivative of the individual log-likelihood function with respect to θk is

51

∂lnPijh∂θk

=1

Pijh· 1

R

R∑r=1

[AjrBr

(Ahr

C−jr

)′+

(AjrBr

)′(Ahr

C−jr

)]

=1

Pijh· 1

R

R∑r=1

{AjrBr

Ahr

[∂δh(θ)∂θk

+ ∂µihr(θ)∂θk

]C−jr − Ahr

∑g 6=j Agr

[∂δg(θ)

∂θk+

∂µigr(θ)

∂θk

](C−jr

)2+Ajr

[∂δj(θ)

∂θk+

∂µijr(θ)

∂θk

]Br − Ajr

∑g Agr

[∂δg(θ)

∂θk+

∂µigr(θ)

∂θk

](Br)

2

Ahr

C−jr

}=

1

Pijh· 1

R

R∑1

{AjrBr

Ahr

C−jr

[∂δh(θ)

∂θk+∂µihr(θ)

∂θk

]− AjrBr

Ahr

C−jr

∑g 6=j

Agr

C−jr

[∂δg(θ)

∂θk+∂µigr(θ)

∂θk

]+AjrBr

Ahr

C−jr

[∂δj(θ)

∂θk+∂µijr(θ)

∂θk

]− AjrBr

Ahr

C−jr

∑g

AgrBr

[∂δg(θ)

∂θk+∂µigr(θ)

∂θk

]}

=1

Pijh· 1

R

R∑1

{Pijhr

[∂δh(θ)

∂θk+∂µihr(θ)

∂θk−∑g 6=j

P−jigr∂δg(θ)

∂θk−∑g 6=j

P−jigr∂µigr(θ)

∂θk

]

+ Pijhr

[∂δj(θ)

∂θk+∂µijr(θ)

∂θk−∑g

Pigr∂δg(θ)

∂θk−∑g

Pigr∂µigr(θ)

∂θk

]},

where Pigr = Agr

Br, which is the probability of i choosing g as first choice conditional on a random

draw r. P−jigr = Agr

B−jr

, which is the probability of i choosing g as a second choice from the choice set

that excludes the first choice j conditional on a random draw r. δj(θ) is estimated via contraction

mapping and does not have an analytical solution. To estimate∂δj(θ)

∂θk, we use implicit function

theorem:

∂δj(θ)

∂θk= −(

∂Sj∂δj

)−1∂Sj∂θk

,

where Sj is the predicted market share for model j, and

∂Sj∂δj

=1

N

N∑i=1

∂Pij∂δj

=1

N

N∑i=1

Pij(1− Pij) =1

N

N∑i=1

Pij −1

N

N∑i=1

P 2ij

∂Sj∂θk

=1

N

N∑i=1

∂Pij∂θk

=1

N

N∑i=1

Pij∂uij∂θk− 1

N

N∑i=1

Pij∑g

Pig∂uig∂θk

.

52

∂δh(θ)∂θk

is estimated by the same method. The score matrix is then:

∂lnR

∂θ=

∂lnP1jh

∂θ1

∂lnP1jh

∂θ2...

∂lnP1jh

∂θk∂lnP2jh

∂θ1

∂lnP2jh

∂θ2...

∂lnP2jh

∂θk

... ... ... ...∂lnPNjh

∂θ1

∂lnPNjh

∂θ2...

∂lnPNjh

∂θk

︸ ︷︷ ︸

N×k

.

Then, averaging the above matrix across households using the sample weight for each

household gives the gradient of the likelihood function, which is the average of the scores across

individuals:

∂lnL

∂θ=[ ∑N

i=1wi ·∂lnPijh

∂θ1

∑Ni=1wi ·

∂lnPijh

∂θ2...

∑Ni=1wi ·

∂lnPijh

∂θk

]︸ ︷︷ ︸

1×k

,

where wi is the sample weight for household i.

E Derivation of Own- and Cross-Price Elasticities of

Demand

The predicted market share for model j is the average of individual probabilities of purchasing

vehicle model j:

sj =1

N

N∑i=1

Pij =1

N

N∑i=1

exp(uij)

1 +∑

g exp(uij).

The partial derivative of individual probability of purchasing j with respect to the price of j

is:

∂Pij∂pj

=exp(uij)

∂uij∂pj

(1 +∑

g exp(uij))− exp(uij)exp(uij)∂uij∂pj

(1 +∑

g exp(uij))2

=exp(uij)

1 +∑J

j=1 exp(uij)

∂uij∂pj−

(exp(uij)

1 +∑J

j=1 exp(uij)

)2∂uij∂pj

= (Pij − P 2ij)∂uij∂pj

.

53

The own-price elasticity of product j is:

∂lnsj∂lnpj

=pjsj· 1

N

N∑i=1

(Pij − P 2ij)∂uij∂pj

.

The partial derivative of individual probability of purchasing j with respect to the price of

model k is:

∂Pij∂pk

=0− exp(uij)exp(uik)∂uik∂pk

(1 +∑J

j=1 exp(uij))2

= −PijPik∂uik∂pk

.

The cross-price elasticity of product j with respect to price of product k is thus:

∂lnsj∂lnpk

=pksj· 1

N

N∑i=1

−PijPik∂uik∂pk

.

F The Effect of Electric Vehicles on Hybrid Sales

One possible concern with promoting electric vehicles is that the policy may have reduced demand

for alternative clean vehicles such as hybrids. Before the introduction of electric vehicles, hybrid

technology was anticipated to provide a pathway to dramatically reducing emissions from light

duty vehicles. Since we have shown that hybrids are relatively close substitutes for electric

vehicles, the introduction of electric vehicles has reduced hybrid market share and may explain

why the market share of hybrids has not continued to grow as it did during the 2000s.

To quantify the impact of electric vehicles on hybrid sales, we conduct several counterfactual

exercises. In scenario 1, we remove all the EV models from the choice sets, and then predict

the counterfactual market shares of the remaining fuel types for MY 2010-MY 2014. In scenario

2, we remove the federal income tax credits for EVs, which vary across EV models, and then

predict the market shares of all fuel types. In panel (a) of Appendix Figure A.3, we plot the

observed and simulated market shares of hybrid vehicles. The dashed lines represent out-of-

sample shares that we obtained from Wards Automotive. The observed market share of hybrids

grew substantially from 2000 to 2010, then eventually leveled off by 2015, around the time that

EVs began to gain a non-trivial market share. As shown in the figure, the simulated market share

54

for hybrids within our sample shows little difference from the observed hybrid market share. We

conclude, therefore, that the introduction of EVs has had only a small impact on the sales of

hybrids during our sample period.

The federal government provided income tax credits of up to $3,400 for hybrid vehicles

purchased after December 31, 2005. All the hybrid vehicles purchased after December 31, 2010,

were not eligible for this credit. The termination of federal support for hybrid vehicles could have

discouraged the sales of hybrid vehicles. To investigate this impact, we conduct simulations to

estimate the counterfactual sales of hybrids if the federal government had continued subsidizing

hybrid vehicles.

We assume that all the hybrid vehicle models that were once subsidized by the government

continue being subsidized with the original subsidy amounts. We run three counterfactual

scenarios (panel (b) of Appendix Figure A.3). In scenario 1, we assume the government continues

subsidizing hybrid vehicles during MY 2010-MY 2014 while also maintaining its subsidy for EVs.

In scenario 2, the government continues subsidizing hybrids during MY 2010-MY 2014 but does

not provide any subsidy for EVs. In scenario 3, the government continues subsidizing hybrids,

but EVs were removed from the market. As implied by the results, the removal of the subsidy

for hybrids played a larger role in reducing the sales of hybrid vehicles, than did the competition

from EVs. Beresteanu and Li (2011) find that the federal income tax credit program for hybrids

contributes to 20 percent of the total sales of hybrid vehicles. This suggests that although the

introduction of EVs has slightly reduced the market share of hybrids, the elimination of the

income tax credit is the major factor in the decline of hybrid vehicle sales in recent years.

55

G Additional Tables and Figures

Table A.1: Sales Shares and Available Models of Hybrids and EVs, 2000-2017

Hybrid EV No. of hybrid No. of EVYears Share Share Models Offered models offered2000 0.05 0.00 2 02001 0.12 0.00 2 02002 0.21 0.00 3 02003 0.29 0.00 3 02004 0.49 0.00 4 02005 1.23 0.00 8 02006 1.52 0.00 10 02007 2.15 0.00 15 02008 2.37 0.00 17 02009 2.77 0.00 21 02010 2.37 0.00 30 22011 2.09 0.14 33 42012 3.01 0.37 44 112013 3.19 0.63 50 172014 2.75 0.72 50 222015 2.21 0.66 51 282016 1.99 0.90 52 332017 2.13 1.14 45 41

Notes: The statistics presented are derived from Wards Automotive new

vehicle sales data, which provide annual estimates of sales by make, model,

and fuel type. Shares are defined as annual sales of the vehicle type divided

by total annual sales. To compute statistics for EVs, we define EV models

that are identified in the Wards data as either plug-in electric or battery

electric vehicles.

56

Table A.2: Vehicle emissions per 100 miles (National average grid mix)

Vehicle Fuel Type GHG EmissionsGasoline 87 lb. CO2

Hybrid Electric 57 lb. CO2

Plug-in Hybrid Electric 62 lb. CO2

All electric 54 lb. CO2

Notes: The data are from the AFDC, https://www.afdc.energy.

gov/vehicles/electric_emissions_sources.html (accessed on

October 15, 2018). The emissions from electric vehicles are

calculated based on the national average fuel sources used in

electricity generation in the United States.

Table A.3: Summary Statistics of Household Data by Fuel Type

EV Buyers Non-EV BuyersMean S.D. Mean S.D.

Household income (1,000$) 180.98 134.73 131.94 107.44With a college degree 0.81 0.40 0.60 0.49Charging stations 0.87 1.76 0.29 0.93Living in an urban area 0.78 0.41 0.64 0.48Average gasoline price ($) 3.73 0.25 3.43 0.40Average vehicle price (1,000$) 32.58 13.28 32.37 12.03Leasing the vehicle 0.44 0.50 0.15 0.36Observations 2,018 9,610

Notes: The data represent the 11,628 observations drawn from the MaritzCX

household survey data. Household income is converted to 2014 dollars using the

BLS calculator. With a college degree indicates whether the purchaser of the

vehicle has obtained a college degree. The number of charging stations is the

cumulative sum of the charging stations that have been built in the survey

respondent’s neighborhood (zip code) by the purchase date of the vehicle.

The charging stations data are collected from the AFDC. Gasoline prices are

monthly average regional prices from the EIA. Average vehicle price is the

average transaction price in year when the vehicle was purchased.

57

Table A.4: Alternative Demand Estimation Results Using MSRPs

Panel (a): Mean Utility Parameters

(1) OLS (2) IVCoefficient S.E. Coefficient S.E.

constant -2.8051 0.1593 -2.8044 0.1609log(price) -1.7856 0.2656 -1.8039 0.9059horsepower/weight 0.702 0.1988 0.6991 0.2413weight 1.8545 0.3467 1.8717 0.8669gallons/mile 0.0663 0.1256 0.0636 0.172AFV dummy -2.5029 0.514 -2.5007 0.5278EV dummy -2.7065 0.2725 -2.7047 0.2889pickup dummy 1.0978 0.3253 1.113 0.7834model year 11 dummy 1.5352 0.1641 1.5436 0.4298model year 12 dummy 1.4235 0.1626 1.4318 0.4264model year 13 dummy 1.1515 0.1645 1.1598 0.421model year 14 dummy 0.0916 0.0894 0.0912 0.0907

Panel (b): Heterogeneous Utility Parameters

Coefficient S.E.Observed Heterogeneitylog(price)/income -7.9676 0.4324family size*vehicle weight 0.1035 0.0204urban*pickups -0.6276 0.0561urban*EV 0.2019 0.0482gasoline price*gallons/mile -0.2957 0.0233education*EV 0.6836 0.0811stations*EV 0.6773 0.1032

Random coefficientsgallons/mile 1.9174 0.0562horsepower/weight 1.11 0.0393light trucks 0.2654 0.0268AFVs 0.9341 0.0892

Average own-price elasticity -2.61Average own-price elasticity of EVs -2.51Average elasticity of EVs (income < $100,000) -2.99

Note: the number of observations are 11628. The value of the simulated log-likelihood at

convergence is -144,184.6 based on 150 Halton draws per household. The table provides

alternative demand estimates that use market suggested retail prices (MSRPs) instead of

transaction prices to construct the price variables. The specifications of the estimation are

the same as Table 3.

58

Table A.5: Sales and Incidence Impacts of Removing EVs

Panel (a): Sales Impact

Fuel types Sales change Percentage Average MPGGasoline 86,114 78.7% 27.2Hybrid 13,167 12.0% 45.1Diesel 2,594 2.4% 27.4FFV 7,574 6.9% 22.0All non-EVs 109,449 100% 28.9

Among gasoline vehicles Sales change Percentagelow mpg (<19) 1,972 2.3%medium mpg (>19 & < 25) 20,409 23.7%high mpg (> 25) 63,733 74.0%

Panel (b): Private Consumer Surplus Impact

Income quintile(1) (2) (3) (4) (5)

Average consumer surplus loss per household ($) -33.41 -42.38 -50.35 -58.8 -73.51

Total consumer surplus loss (million $) -670.3

Notes: The table summarizes the sales impact of removing all EVs from the choice set in

MY 2014 on other vehicles of different fuel types and its welfare impact on consumers. The

percentage column in panel (a) reports the percentage of the total sales increase from non-EV

models that each fuel type contributes to. Average mpg represents the average miles per gallon

of the vehicles that experience sales increases as a result of the removal of EVs. In panel

(b), the average consumer surplus loss per household is calculated as the change in consumer

surplus as shown by Small and Rosen (1981) for the year 2014. Total consumer surplus loss

is calculated as average consumer surplus loss multiplied by the market size of new vehicles.

Consumer surplus calculation does not include welfare impacts of emissions.

59

Table A.6: Predicted Lifetime Vehicle Miles Traveled

Segment Gasoline Diesel Hybrid Plug-in Hybrid All Eletric Lifetime VMTCar 0 0 0 0 1 136,328Car 0 0 0 1 0 145,589Car 0 0 1 0 0 155,611Car 0 1 0 0 0 178,898Car 1 0 0 0 0 147,962Pick Up Truck 0 0 1 0 0 151,179Pick Up Truck 0 1 0 0 0 191,380Pick Up Truck 1 0 0 0 0 173,794SUV 0 0 0 0 1 156,777SUV 0 0 0 1 0 167,199SUV 0 0 1 0 0 176,789SUV 0 1 0 0 0 187,306SUV 1 0 0 0 0 177,850Van 0 0 0 0 1 164,323Van 0 0 0 1 0 163,531Van 0 0 1 0 0 151,910Van 0 1 0 0 0 180,489Van 1 0 0 0 0 162,732

Notes: The table summarizes the predicted segment-fuel-type-specific lifetime vehicle miles traveled

estimated using odometer data following Linn (2020). The detailed procedures are explained in Appendix

Section C.

Table A.7: Environmental Benefits of EV Subsidy

Pollutants Reduction (tons) Damage ($/ton) Damage reduction (million $)CO2 912,019.30 36.0 32.8VOC 2,549.7 1,482.0 3.8NOx 1,710.0 6,042.0 10.3PM2.5 10.2 330,600.0 3.4SO2 17.4 35,340.0 0.6All 50.9

Notes: The table summarizes the environmental benefits of the federal-level income tax credit

for EVs, with a total spending of $725.7 million in 2014. The environmental estimates are based

on the external cost savings from emissions reduction of various pollutants due to less petroleum

consumption. The emissions rates and associated damage values are obtained from EPA (2008).

60

Table A.8: Simulation Results with Parameters Estimated without Second Choice Data

Panel (a): Sales Impact of Removing EVs

Fuel types Sales change Percentage Average MPGGasoline 87,340 79.8% 28.1Hybrid 14,009 12.8% 46.0Diesel 2,846 2.6% 31.8FFV 5,254 4.8% 23.1All non-EVs 109,449 100% 30.3

Among gasoline vehicles Sales change Percentagelow mpg (<19) 678 0.7%medium mpg (>19 & < 25) 13,767 15.8%high mpg (> 25) 72,895 83.5%

Panel (b): Sales Impact of Removing EV Subsidy

Fuel types Sales change Percentage changeEV -27,653 -25.3%BEV -16,518 -28.6%PHEV -11,136 -21.6%Other fuel types Sales change Percentage Change Percentage of EV sales reduction Average MPGGasoline 22,245 0.21% 80.4% 28.2Hybrid 3,379 0.83% 12.2% 46.0Diesel 706 0.14% 2.6% 31.8FFV 1,323 0.15% 4.8% 23.2All non-EVs 27,653 0.22% 100% 30.2

Notes: The table summarizes the sales impact of removing all EVs from the choice set in MY 2014 and the sales impact of

removing the federal EV tax credits in 2014. The simulations are conducted with the estimated parameters without using the

second choice data, which are reported in Table 3.

61

Figure A.1: Consumer Second Choices by Vehicle Segment

(a) Gasoline vehicle buyers (b) Hybrid vehicle buyers

(c) PHEV buyers (d) BEV buyers

Notes: The figure plots the frequency of alternative vehicle choices by vehicle segment for different groups of consumers based on the survey

responses of the 11,628 households in the sample. Light trucks include SUVs and Pickups. The number of observations for the buyers of gasoline

vehicles, hybrid vehicles, PHEVs and BEVs is 9,295, 315, 1,246, and 772, respectively.

62

Figure A.2: Own-price Elasticity Estimates

Notes: The figure plots each vehicle model’s own-price elasticity against its average transaction price in 2014.

The elasticity estimates are calculated based on the parameter estimates reported in Table 6, and are calculated

with the same individual weights and Halton draws used in the demand estimation.

63

Figure A.3: The Effect of Electric Vehicles on Hybrid Sales

Panel (a) No subsidy for hybrids

Panel (b) With subsidy for hybrids

Notes: the figure plots the impact of the introduction of EVs and EV subsidy on

the market shares of hybrid vehicles. The dashed lines represent out-of-sample shares

obtained from Wards Automotive. In Panel (a), Scenario 1 removes EV models from

the market. Scenario 2 removes the federal income tax credits for EVs. In Panel (b), we

assume that the government continues subsidizing hybrid vehicles between 2010-14 with

their original subsidy amount in all the three scenarios. Scenario 1 keeps the current EV

subsidy. Scenario 2 removes EV subsidy and Scenario 3 removes EVs from the market.

64